Tag Archives: Research

Can large language models identify and correct their mistakes?

Posted by Gladys Tyen, Intern, Google Research

LLMs are increasingly popular for reasoning tasks, such as multi-turn QA, task completion, code generation, or mathematics. Yet much like people, they do not always solve problems correctly on the first try, especially on tasks for which they were not trained. Therefore, for such systems to be most useful, they should be able to 1) identify where their reasoning went wrong and 2) backtrack to find another solution.

This has led to a surge in methods related to self-correction, where an LLM is used to identify problems in its own output, and then produce improved results based on the feedback. Self-correction is generally thought of as a single process, but we decided to break it down into two components, mistake finding and output correction.

In “LLMs cannot find reasoning errors, but can correct them!”, we test state-of-the-art LLMs on mistake finding and output correction separately. We present BIG-Bench Mistake, an evaluation benchmark dataset for mistake identification, which we use to address the following questions:

  1. Can LLMs find logical mistakes in Chain-of-Thought (CoT) style reasoning?
  2. Can mistake-finding be used as a proxy for correctness?
  3. Knowing where the mistake is, can LLMs then be prompted to backtrack and arrive at the correct answer?
  4. Can mistake finding as a skill generalize to tasks the LLMs have never seen?

About our dataset

Mistake finding is an underexplored problem in natural language processing, with a particular lack of evaluation tasks in this domain. To best assess the ability of LLMs to find mistakes, evaluation tasks should exhibit mistakes that are non-ambiguous. To our knowledge, most current mistake-finding datasets do not go beyond the realm of mathematics for this reason.

To assess the ability of LLMs to reason about mistakes outside of the math domain, we produce a new dataset for use by the research community, called BIG-Bench Mistake. This dataset consists of Chain-of-Thought traces generated using PaLM 2 on five tasks in BIG-Bench. Each trace is annotated with the location of the first logical mistake.

To maximize the number of mistakes in our dataset, we sample 255 traces where the answer is incorrect (so we know there is definitely a mistake), and 45 traces where the answer is correct (so there may or may not be a mistake). We then ask human labelers to go through each trace and identify the first mistake step. Each trace has been annotated by at least three labelers, whose answers had inter-rater reliability levels of >0.98 (using Krippendorff’s α). The labeling was done for all tasks except the Dyck Languages task, which involves predicting the sequence of closing parentheses for a given input sequence. This task we labeled algorithmically.

The logical errors made in this dataset are simple and unambiguous, providing a good benchmark for testing an LLM’s ability to find its own mistakes before using them on harder, more ambiguous tasks.


Core questions about mistake identification


1. Can LLMs find logical mistakes in Chain-of-Thought style reasoning?

First, we want to find out if LLMs can identify mistakes independently of their ability to correct them. We attempt multiple prompting methods to test GPT series models for their ability to locate mistakes (prompts here) under the assumption that they are generally representative of modern LLM performance.

Generally, we found these state-of-the-art models perform poorly, with the best model achieving 52.9% accuracy overall. Hence, there is a need to improve LLMs’ ability in this area of reasoning.

In our experiments, we try three different prompting methods: direct (trace), direct (step) and CoT (step). In direct (trace), we provide the LLM with the trace and ask for the location step of the mistake or no mistake. In direct (step), we prompt the LLM to ask itself this question for each step it takes. In CoT (step), we prompt the LLM to give its reasoning for whether each step is a mistake or not a mistake.

A diagram showing the three prompting methods direct (trace), direct (step) and CoT (step).

Our finding is in line and builds upon prior results, but goes further in showing that LLMs struggle with even simple and unambiguous mistakes (for comparison, our human raters without prior expertise solve the problem with a high degree of agreement). We hypothesize that this is a big reason why LLMs are unable to self-correct reasoning errors. See the paper for the full results.


2. Can mistake-finding be used as a proxy for correctness of the answer?

When people are confronted with a problem where we are unsure of the answer, we can work through our solutions step-by-step. If no error is found, we can make the assumption that we did the right thing.

While we hypothesized that this would work similarly for LLMs, we discovered that this is a poor strategy. On our dataset of 85% incorrect traces and 15% correct traces, using this method is not much better than the naïve strategy of always labeling traces as incorrect, which gives a weighted average F1 of 78.

A diagram showing how well mistake-finding with LLMs can be used as a proxy for correctness of the answer on each dataset.

3. Can LLMs backtrack knowing where the error is?

Since we’ve shown that LLMs exhibit poor performance in finding reasoning errors in CoT traces, we want to know whether LLMs can even correct errors at all, even if they know where the error is.

Note that knowing the mistake location is different from knowing the right answer: CoT traces can contain logical mistakes even if the final answer is correct, or vice versa. In most real-world situations, we won’t know what the right answer is, but we might be able to identify logical errors in intermediate steps.

We propose the following backtracking method:

  1. Generate CoT traces as usual, at temperature = 0. (Temperature is a parameter that controls the randomness of generated responses, with higher values producing more diverse and creative outputs, usually at the expense of quality.)
  2. Identify the location of the first logical mistake (for example with a classifier, or here we just use labels from our dataset).
  3. Re-generate the mistake step at temperature = 1 and produce a set of eight outputs. Since the original output is known to lead to incorrect results, the goal is to find an alternative generation at this step that is significantly different from the original.
  4. From these eight outputs, select one that is different from the original mistake step. (We just use exact matching here, but in the future this can be something more sophisticated.)
  5. Using the new step, generate the rest of the trace as normal at temperature = 0.

It’s a very simple method that does not require any additional prompt crafting and avoids having to re-generate the entire trace. We test it using the mistake location data from BIG-Bench Mistake, and we find that it can correct CoT errors.

Recent work showed that self-correction methods, like Reflexion and RCI, cause deterioration in accuracy scores because there are more correct answers becoming incorrect than vice versa. Our method, on the other hand, produces more gains (by correcting wrong answers) than losses (by changing right answers to wrong answers).

We also compare our method with a random baseline, where we randomly assume a step to be a mistake. Our results show that this random baseline does produce some gains, but not as much as backtracking with the correct mistake location, and with more losses.

A diagram showing the gains and losses in accuracy for our method as well as a random baseline on each dataset.

4. Can mistake finding generalize to tasks the LLMs have never seen?

To answer this question, we fine-tuned a small model on four of the BIG-Bench tasks and tested it on the fifth, held-out task. We do this for every task, producing five fine-tuned models in total. Then we compare the results with just zero-shot prompting PaLM 2-L-Unicorn, a much larger model.

Bar chart showing the accuracy improvement of the fine-tuned small model compared to zero-shot prompting with PaLM 2-L-Unicorn.

Our results show that the much smaller fine-tuned reward model generally performs better than zero-shot prompting a large model, even though the reward model has never seen data from the task in the test set. The only exception is logical deduction, where it performs on par with zero-shot prompting.

This is a very promising result as we can potentially just use a small fine-tuned reward model to perform backtracking and improve accuracy on any task, even if we don’t have the data for it. This smaller reward model is completely independent of the generator LLM, and can be updated and further fine-tuned for individual use cases.

An illustration showing how our backtracking method works.

Conclusion

In this work, we created an evaluation benchmark dataset that the wider academic community can use to evaluate future LLMs. We further showed that LLMs currently struggle to find logical errors. However, if they could, we show the effectiveness of backtracking as a strategy that can provide gains on tasks. Finally, a smaller reward model can be trained on general mistake-finding tasks and be used to improve out-of-domain mistake finding, showing that mistake-finding can generalize.


Acknowledgements

Thank you to Peter Chen, Tony Mak, Hassan Mansoor and Victor Cărbune for contributing ideas and helping with the experiments and data collection. We would also like to thank Sian Gooding and Vicky Zayats for their comments and suggestions on the paper.


Source: Google AI Blog


Responsible AI at Google Research: User Experience Team

Posted by Ayça Çakmakli, UX Lead, Google Research, Responsible AI and Human Centered Technology Team

Google’s Responsible AI User Experience (Responsible AI UX) team is a product-minded team embedded within Google Research. This unique positioning requires us to apply responsible AI development practices to our user-centered user experience (UX) design process. In this post, we describe the importance of UX design and responsible AI in product development, and share a few examples of how our team’s capabilities and cross-functional collaborations have led to responsible development across Google.

First, the UX part. We are a multi-disciplinary team of product design experts: designers, engineers, researchers, and strategists who manage the user-centered UX design process from early-phase ideation and problem framing to later-phase user-interface (UI) design, prototyping and refinement. We believe that effective product development occurs when there is clear alignment between significant unmet user needs and a product's primary value proposition, and that this alignment is reliably achieved via a thorough user-centered UX design process.

And second, recognizing generative AI’s (GenAI) potential to significantly impact society, we embrace our role as the primary user advocate as we continue to evolve our UX design process to meet the unique challenges AI poses, maximizing the benefits and minimizing the risks. As we navigate through each stage of an AI-powered product design process, we place a heightened emphasis on the ethical, societal, and long-term impact of our decisions. We contribute to the ongoing development of comprehensive safety and inclusivity protocols that define design and deployment guardrails around key issues like content curation, security, privacy, model capabilities, model access, equitability, and fairness that help mitigate GenAI risks.

Responsible AI UX is constantly evolving its user-centered product design process to meet the needs of a GenAI-powered product landscape with greater sensitivity to the needs of users and society and an emphasis on ethical, societal, and long-term impact.

Responsibility in product design is also reflected in the user and societal problems we choose to address and the programs we resource. Thus, we encourage the prioritization of user problems with significant scale and severity to help maximize the positive impact of GenAI technology.

Communication across teams and disciplines is essential to responsible product design. The seamless flow of information and insight from user research teams to product design and engineering teams, and vice versa, is essential to good product development. One of our team’s core objectives is to ensure the practical application of deep user-insight into AI-powered product design decisions at Google by bridging the communication gap between the vast technological expertise of our engineers and the user/societal expertise of our academics, research scientists, and user-centered design research experts. We’ve built a multidisciplinary team with expertise in these areas, deepening our empathy for the communication needs of our audience, and enabling us to better interface between our user & society experts and our technical experts. We create frameworks, guidebooks, prototypes, cheatsheets, and multimedia tools to help bring insights to life for the right people at the right time.



Facilitating responsible GenAI prototyping and development

During collaborations between Responsible AI UX, the People + AI Research (PAIR) initiative and Labs, we identified that prototyping can afford a creative opportunity to engage with large language models (LLM), and is often the first step in GenAI product development. To address the need to introduce LLMs into the prototyping process, we explored a range of different prompting designs. Then, we went out into the field, employing various external, first-person UX design research methodologies to draw out insight and gain empathy for the user’s perspective. Through user/designer co-creation sessions, iteration, and prototyping, we were able to bring internal stakeholders, product managers, engineers, writers, sales, and marketing teams along to ensure that the user point of view was well understood and to reinforce alignment across teams.

The result of this work was MakerSuite, a generative AI platform launched at Google I/O 2023 that enables people, even those without any ML experience, to prototype creatively using LLMs. The team’s first-hand experience with users and understanding of the challenges they face allowed us to incorporate our AI Principles into the MakerSuite product design. Product features like safety filters, for example, enable users to manage outcomes, leading to easier and more responsible product development with MakerSuite.

Because of our close collaboration with product teams, we were able to adapt text-only prototyping to support multimodal interaction with Google AI Studio, an evolution of MakerSuite. Now, Google AI Studio enables developers and non-developers alike to seamlessly leverage Google’s latest Gemini model to merge multiple modality inputs, like text and image, in product explorations. Facilitating product development in this way provides us with the opportunity to better use AI to identify appropriateness of outcomes and unlocks opportunities for developers and non-developers to play with AI sandboxes. Together with our partners, we continue to actively push this effort in the products we support.

Google AI studio enables developers and non-developers to leverage Google Cloud infrastructure and merge multiple modality inputs in their product explorations.


Equitable speech recognition

Multiple external studies, as well as Google’s own research, have identified an unfortunate deficiency in the ability of current speech recognition technology to understand Black speakers on average, relative to White speakers. As multimodal AI tools begin to rely more heavily on speech prompts, this problem will grow and continue to alienate users. To address this problem, the Responsible AI UX team is partnering with world-renowned linguists and scientists at Howard University, a prominent HBCU, to build a high quality African-American English dataset to improve the design of our speech technology products to make them more accessible. Called Project Elevate Black Voices, this effort will allow Howard University to share the dataset with those looking to improve speech technology while establishing a framework for responsible data collection, ensuring the data benefits Black communities. Howard University will retain the ownership and licensing of the dataset and serve as stewards for its responsible use. At Google, we’re providing funding support and collaborating closely with our partners at Howard University to ensure the success of this program.




Equitable computer vision

The Gender Shades project highlighted that computer vision systems struggle to detect people with darker skin tones, and performed particularly poorly for women with darker skin tones. This is largely due to the fact that the datasets used to train these models were not inclusive to a wide range of skin tones. To address this limitation, the Responsible AI UX team has been partnering with sociologist Dr. Ellis Monk to release the Monk Skin Tone Scale (MST), a skin tone scale designed to be more inclusive of the spectrum of skin tones around the world. It provides a tool to assess the inclusivity of datasets and model performance across an inclusive range of skin tones, resulting in features and products that work better for everyone.

We have integrated MST into a range of Google products, such as Search, Google Photos, and others. We also open sourced MST, published our research, described our annotation practices, and shared an example dataset to encourage others to easily integrate it into their products. The Responsible AI UX team continues to collaborate with Dr. Monk, utilizing the MST across multiple product applications and continuing to do international research to ensure that it is globally inclusive.


Consulting & guidance

As teams across Google continue to develop products that leverage the capabilities of GenAI models, our team recognizes that the challenges they face are varied and that market competition is significant. To support teams, we develop actionable assets to facilitate a more streamlined and responsible product design process that considers available resources. We act as a product-focused design consultancy, identifying ways to scale services, share expertise, and apply our design principles more broadley. Our goal is to help all product teams at Google connect significant unmet user needs with technology benefits via great responsible product design.

One way we have been doing this is with the creation of the People + AI Guidebook, an evolving summative resource of many of the responsible design lessons we’ve learned and recommendations we’ve made for internal and external stakeholders. With its forthcoming, rolling updates focusing specifically on how to best design and consider user needs with GenAI, we hope that our internal teams, external stakeholders, and larger community will have useful and actionable guidance at the most critical milestones in the product development journey.

The People + AI Guidebook has six chapters, designed to cover different aspects of the product life cycle.

If you are interested in reading more about Responsible AI UX and how we are specifically thinking about designing responsibly with Generative AI, please check out this Q&A piece.


Acknowledgements

Shout out to our the Responsible AI UX team members: Aaron Donsbach, Alejandra Molina, Courtney Heldreth, Diana Akrong, Ellis Monk, Femi Olanubi, Hope Neveux, Kafayat Abdul, Key Lee, Mahima Pushkarna, Sally Limb, Sarah Post, Sures Kumar Thoddu Srinivasan, Tesh Goyal, Ursula Lauriston, and Zion Mengesha. Special thanks to Michelle Cohn for her contributions to this work.

Source: Google AI Blog


2023: A year of groundbreaking advances in AI and computing

Posted by Jeff Dean, Chief Scientist, Google DeepMind & Google Research, Demis Hassabis, CEO, Google DeepMind, and James Manyika, SVP, Google Research, Technology & Society

This has been a year of incredible progress in the field of Artificial Intelligence (AI) research and its practical applications.

As ongoing research pushes AI even farther, we look back to our perspective published in January of this year, titled “Why we focus on AI (and to what end),” where we noted:

We are committed to leading and setting the standard in developing and shipping useful and beneficial applications, applying ethical principles grounded in human values, and evolving our approaches as we learn from research, experience, users, and the wider community.

We also believe that getting AI right — which to us involves innovating and delivering widely accessible benefits to people and society, while mitigating its risks — must be a collective effort involving us and others, including researchers, developers, users (individuals, businesses, and other organizations), governments, regulators, and citizens.

We are convinced that the AI-enabled innovations we are focused on developing and delivering boldly and responsibly are useful, compelling, and have the potential to assist and improve lives of people everywhere — this is what compels us.

In this Year-in-Review post we’ll go over some of Google Research's and Google DeepMind’s efforts putting these paragraphs into practice safely throughout 2023.


Advances in products & technologies

This was the year generative AI captured the world’s attention, creating imagery, music, stories, and engaging conversation about everything imaginable, at a level of creativity and a speed almost implausible a few years ago.

In February, we first launched Bard, a tool that you can use to explore creative ideas and explain things simply. It can generate text, translate languages, write different kinds of creative content and more.

In May, we watched the results of months and years of our foundational and applied work announced on stage at Google I/O. Principally, this included PaLM 2, a large language model (LLM) that brought together compute-optimal scaling, an improved dataset mixture, and model architecture to excel at advanced reasoning tasks.

By fine-tuning and instruction-tuning PaLM 2 for different purposes, we were able to integrate it into numerous Google products and features, including:

  • An update to Bard, which enabled multilingual capabilities. Since its initial launch, Bard is now available in more than 40 languages and over 230 countries and territories, and with extensions, Bard can find and show relevant information from Google tools used every day — like Gmail, Google Maps, YouTube, and more.
  • Search Generative Experience (SGE), which uses LLMs to reimagine both how to organize information and how to help people navigate through it, creating a more fluid, conversational interaction model for our core Search product. This work extended the search engine experience from primarily focused on information retrieval into something much more — capable of retrieval, synthesis, creative generation and continuation of previous searches — while continuing to serve as a connection point between users and the web content they seek.
  • MusicLM, a text-to-music model powered by AudioLM and MuLAN, which can make music from text, humming, images or video and musical accompaniments to singing.
  • Duet AI, our AI-powered collaborator that provides users with assistance when they use Google Workspace and Google Cloud. Duet AI in Google Workspace, for example, helps users write, create images, analyze spreadsheets, draft and summarize emails and chat messages, and summarize meetings. Duet AI in Google Cloud helps users code, deploy, scale, and monitor applications, as well as identify and accelerate resolution of cybersecurity threats.
  • And many other developments.

In June, following last year’s release of our text-to-image generation model Imagen, we released Imagen Editor, which provides the ability to use region masks and natural language prompts to interactively edit generative images to provide much more precise control over the model output.

Later in the year, we released Imagen 2, which improved outputs via a specialized image aesthetics model based on human preferences for qualities such as good lighting, framing, exposure, and sharpness.

In October, we launched a feature that helps people practice speaking and improve their language skills. The key technology that enabled this functionality was a novel deep learning model developed in collaboration with the Google Translate team, called Deep Aligner. This single new model has led to dramatic improvements in alignment quality across all tested language pairs, reducing average alignment error rate from 25% to 5% compared to alignment approaches based on Hidden Markov models (HMMs).

In November, in partnership with YouTube, we announced Lyria, our most advanced AI music generation model to date. We released two experiments designed to open a new playground for creativity, DreamTrack and music AI tools, in concert with YouTube’s Principles for partnering with the music industry on AI technology.

Then in December, we launched Gemini, our most capable and general AI model. Gemini was built to be multimodal from the ground up across text, audio, image and videos. Our initial family of Gemini models comes in three different sizes, Nano, Pro, and Ultra. Nano models are our smallest and most efficient models for powering on-device experiences in products like Pixel. The Pro model is highly-capable and best for scaling across a wide range of tasks. The Ultra model is our largest and most capable model for highly complex tasks.



In a technical report about Gemini models, we showed that Gemini Ultra’s performance exceeds current state-of-the-art results on 30 of the 32 widely-used academic benchmarks used in LLM research and development. With a score of 90.04%, Gemini Ultra was the first model to outperform human experts on MMLU, and achieved a state-of-the-art score of 59.4% on the new MMMU benchmark.

Building on AlphaCode, the first AI system to perform at the level of the median competitor in competitive programming, we introduced AlphaCode 2 powered by a specialized version of Gemini. When evaluated on the same platform as the original AlphaCode, we found that AlphaCode 2 solved 1.7x more problems, and performed better than 85% of competition participants

At the same time, Bard got its biggest upgrade with its use of the Gemini Pro model, making it far more capable at things like understanding, summarizing, reasoning, coding, and planning. In six out of eight benchmarks, Gemini Pro outperformed GPT-3.5, including in MMLU, one of the key standards for measuring large AI models, and GSM8K, which measures grade school math reasoning. Gemini Ultra will come to Bard early next year through Bard Advanced, a new cutting-edge AI experience.

Gemini Pro is also available on Vertex AI, Google Cloud’s end-to-end AI platform that empowers developers to build applications that can process information across text, code, images, and video. Gemini Pro was also made available in AI Studio in December.

To best illustrate some of Gemini’s capabilities, we produced a series of short videos with explanations of how Gemini could:


ML/AI Research

In addition to our advances in products and technologies, we’ve also made a number of important advancements in the broader fields of machine learning and AI research.

At the heart of the most advanced ML models is the Transformer model architecture, developed by Google researchers in 2017. Originally developed for language, it has proven useful in domains as varied as computer vision, audio, genomics, protein folding, and more. This year, our work on scaling vision transformers demonstrated state-of-the-art results across a wide variety of vision tasks, and has also been useful in building more capable robots.

Expanding the versatility of models requires the ability to perform higher-level and multi-step reasoning. This year, we approached this target following several research tracks. For example, algorithmic prompting is a new method that teaches language models reasoning by demonstrating a sequence of algorithmic steps, which the model can then apply in new contexts. This approach improves accuracy on one middle-school mathematics benchmark from 25.9% to 61.1%.

By providing algorithmic prompts, we can teach a model the rules of arithmetic via in-context learning.

In the domain of visual question answering, in a collaboration with UC Berkeley researchers, we showed how we could better answer complex visual questions (“Is the carriage to the right of the horse?”) by combining a visual model with a language model trained to answer visual questions by synthesizing a program to perform multi-step reasoning.

We are now using a general model that understands many aspects of the software development life cycle to automatically generate code review comments, respond to code review comments, make performance-improving suggestions for pieces of code (by learning from past such changes in other contexts), fix code in response to compilation errors, and more.

In a multi-year research collaboration with the Google Maps team, we were able to scale inverse reinforcement learning and apply it to the world-scale problem of improving route suggestions for over 1 billion users. Our work culminated in a 16–24% relative improvement in global route match rate, helping to ensure that routes are better aligned with user preferences.

We also continue to work on techniques to improve the inference performance of machine learning models. In work on computationally-friendly approaches to pruning connections in neural networks, we were able to devise an approximation algorithm to the computationally intractable best-subset selection problem that is able to prune 70% of the edges from an image classification model and still retain almost all of the accuracy of the original.

In work on accelerating on-device diffusion models, we were also able to apply a variety of optimizations to attention mechanisms, convolutional kernels, and fusion of operations to make it practical to run high quality image generation models on-device; for example, enabling “a photorealistic and high-resolution image of a cute puppy with surrounding flowers” to be generated in just 12 seconds on a smartphone.



Advances in capable language and multimodal models have also benefited our robotics research efforts. We combined separately trained language, vision, and robotic control models into PaLM-E, an embodied multi-modal model for robotics, and Robotic Transformer 2 (RT-2), a novel vision-language-action (VLA) model that learns from both web and robotics data, and translates this knowledge into generalized instructions for robotic control.

RT-2 architecture and training: We co-fine-tune a pre-trained vision-language model on robotics and web data. The resulting model takes in robot camera images and directly predicts actions for a robot to perform.

Furthermore, we showed how language can also be used to control the gait of quadrupedal robots and explored the use of language to help formulate more explicit reward functions to bridge the gap between human language and robotic actions. Then, in Barkour we benchmarked the agility limits of quadrupedal robots.


Algorithms & optimization

Designing efficient, robust, and scalable algorithms remains a high priority. This year, our work included: applied and scalable algorithms, market algorithms, system efficiency and optimization, and privacy.

We introduced AlphaDev, an AI system that uses reinforcement learning to discover enhanced computer science algorithms. AlphaDev uncovered a faster algorithm for sorting, a method for ordering data, which led to improvements in the LLVM libc++ sorting library that were up to 70% faster for shorter sequences and about 1.7% faster for sequences exceeding 250,000 elements.

We developed a novel model to predict the properties of large graphs, enabling estimation of performance for large programs. We released a new dataset, TPUGraphs, to accelerate open research in this area, and showed how we can use modern ML to improve ML efficiency.

The TPUGraphs dataset has 44 million graphs for ML program optimization.

We developed a new load balancing algorithm for distributing queries to a server, called Prequal, which minimizes a combination of requests-in-flight and estimates the latency. Deployments across several systems have saved CPU, latency, and RAM significantly. We also designed a new analysis framework for the classical caching problem with capacity reservations.

Heatmaps of normalized CPU usage transitioning to Prequal at 08:00.

We improved state-of-the-art in clustering and graph algorithms by developing new techniques for computing minimum-cut, approximating correlation clustering, and massively parallel graph clustering. Additionally, we introduced TeraHAC, a novel hierarchical clustering algorithm for trillion-edge graphs, designed a text clustering algorithm for better scalability while maintaining quality, and designed the most efficient algorithm for approximating the Chamfer Distance, the standard similarity function for multi-embedding models, offering >50× speedups over highly-optimized exact algorithms and scaling to billions of points.

We continued optimizing Google’s large embedding models (LEMs), which power many of our core products and recommender systems. Some new techniques include Unified Embedding for battle-tested feature representations in web-scale ML systems and Sequential Attention, which uses attention mechanisms to discover high-quality sparse model architectures during training.

Beyond auto-bidding systems, we also studied auction design in other complex settings, such as buy-many mechanisms, auctions for heterogeneous bidders, contract designs, and innovated robust online bidding algorithms. Motivated by the application of generative AI in collaborative creation (e.g., joint ad for advertisers), we proposed a novel token auction model where LLMs bid for influence in the collaborative AI creation. Finally, we show how to mitigate personalization effects in experimental design, which, for example, may cause recommendations to drift over time.

The Chrome Privacy Sandbox, a multi-year collaboration between Google Research and Chrome, has publicly launched several APIs, including for Protected Audience, Topics, and Attribution Reporting. This is a major step in protecting user privacy while supporting the open and free web ecosystem. These efforts have been facilitated by fundamental research on re-identification risk, private streaming computation, optimization of privacy caps and budgets, hierarchical aggregation, and training models with label privacy.


Science and society

In the not too distant future, there is a very real possibility that AI applied to scientific problems can accelerate the rate of discovery in certain domains by 10× or 100×, or more, and lead to major advances in diverse areas including bioengineering, materials science, weather prediction, climate forecasting, neuroscience, genetic medicine, and healthcare.


Sustainability and climate change

In Project Green Light, we partnered with 13 cities around the world to help improve traffic flow at intersections and reduce stop-and-go emissions. Early numbers from these partnerships indicate a potential for up to 30% reduction in stops and up to 10% reduction in emissions.

In our contrails work, we analyzed large-scale weather data, historical satellite images, and past flights. We trained an AI model to predict where contrails form and reroute airplanes accordingly. In partnership with American Airlines and Breakthrough Energy, we used this system to demonstrate contrail reduction by 54%.

Contrails detected over the United States using AI and GOES-16 satellite imagery.

We are also developing novel technology-driven approaches to help communities with the effects of climate change. For example, we have expanded our flood forecasting coverage to 80 countries, which directly impacts more than 460 million people. We have initiated a number of research efforts to help mitigate the increasing danger of wildfires, including real-time tracking of wildfire boundaries using satellite imagery, and work that improves emergency evacuation plans for communities at risk to rapidly-spreading wildfires. Our partnership with American Forests puts data from our Tree Canopy project to work in their Tree Equity Score platform, helping communities identify and address unequal access to trees.

Finally, we continued to develop better models for weather prediction at longer time horizons. Improving on MetNet and MetNet-2, in this year’s work on MetNet-3, we now outperform traditional numerical weather simulations up to twenty-four hours. In the area of medium-term, global weather forecasting, our work on GraphCast showed significantly better prediction accuracy for up to 10 days compared to HRES, the most accurate operational deterministic forecast, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). In collaboration with ECMWF, we released WeatherBench-2, a benchmark for evaluating the accuracy of weather forecasts in a common framework.


A selection of GraphCast’s predictions rolling across 10 days showing specific humidity at 700 hectopascals (about 3 km above surface), surface temperature, and surface wind speed.

Health and the life sciences

The potential of AI to dramatically improve processes in healthcare is significant. Our initial Med-PaLM model was the first model capable of achieving a passing score on the U.S. medical licensing exam. Our more recent Med-PaLM 2 model improved by a further 19%, achieving an expert-level accuracy of 86.5%. These Med-PaLM models are language-based, enable clinicians to ask questions and have a dialogue about complex medical conditions, and are available to healthcare organizations as part of MedLM through Google Cloud.

In the same way our general language models are evolving to handle multiple modalities, we have recently shown research on a multimodal version of Med-PaLM capable of interpreting medical images, textual data, and other modalities, describing a path for how we can realize the exciting potential of AI models to help advance real-world clinical care.

Med-PaLM M is a large multimodal generative model that flexibly encodes and interprets biomedical data including clinical language, imaging, and genomics with the same model weights.

We have also been working on how best to harness AI models in clinical workflows. We have shown that coupling deep learning with interpretability methods can yield new insights for clinicians. We have also shown that self-supervised learning, with careful consideration of privacy, safety, fairness and ethics, can reduce the amount of de-identified data needed to train clinically relevant medical imaging models by 3×–100×, reducing the barriers to adoption of models in real clinical settings. We also released an open source mobile data collection platform for people with chronic disease to provide tools to the community to build their own studies.

AI systems can also discover completely new signals and biomarkers in existing forms of medical data. In work on novel biomarkers discovered in retinal images, we demonstrated that a number of systemic biomarkers spanning several organ systems (e.g., kidney, blood, liver) can be predicted from external eye photos. In other work, we showed that combining retinal images and genomic information helps identify some underlying factors of aging.

In the genomics space, we worked with 119 scientists across 60 institutions to create a new map of the human genome, or pangenome. This more equitable pangenome better represents the genomic diversity of global populations. Building on our ground-breaking AlphaFold work, our work on AlphaMissense this year provides a catalog of predictions for 89% of all 71 million possible missense variants as either likely pathogenic or likely benign.

Examples of AlphaMissense predictions overlaid on AlphaFold predicted structures (red – predicted as pathogenic; blue – predicted as benign; grey – uncertain). Red dots represent known pathogenic missense variants, blue dots represent known benign variants. Left: HBB protein. Variants in this protein can cause sickle cell anaemia. Right: CFTR protein. Variants in this protein can cause cystic fibrosis.

We also shared an update on progress towards the next generation of AlphaFold. Our latest model can now generate predictions for nearly all molecules in the Protein Data Bank (PDB), frequently reaching atomic accuracy. This unlocks new understanding and significantly improves accuracy in multiple key biomolecule classes, including ligands (small molecules), proteins, nucleic acids (DNA and RNA), and those containing post-translational modifications (PTMs).

On the neuroscience front, we announced a new collaboration with Harvard, Princeton, the NIH, and others to map an entire mouse brain at synaptic resolution, beginning with a first phase that will focus on the hippocampal formation — the area of the brain responsible for memory formation, spatial navigation, and other important functions.


Quantum computing

Quantum computers have the potential to solve big, real-world problems across science and industry. But to realize that potential, they must be significantly larger than they are today, and they must reliably perform tasks that cannot be performed on classical computers.

This year, we took an important step towards the development of a large-scale, useful quantum computer. Our breakthrough is the first demonstration of quantum error correction, showing that it’s possible to reduce errors while also increasing the number of qubits. To enable real-world applications, these qubit building blocks must perform more reliably, lowering the error rate from ~1 in 103 typically seen today, to ~1 in 108.


Responsible AI research


Design for Responsibility

Generative AI is having a transformative impact in a wide range of fields including healthcare, education, security, energy, transportation, manufacturing, and entertainment. Given these advances, the importance of designing technologies consistent with our AI Principles remains a top priority. We also recently published case studies of emerging practices in society-centered AI. And in our annual AI Principles Progress Update, we offer details on how our Responsible AI research is integrated into products and risk management processes.

Proactive design for Responsible AI begins with identifying and documenting potential harms. For example, we recently introduced a three-layered context-based framework for comprehensively evaluating the social and ethical risks of AI systems. During model design, harms can be mitigated with the use of responsible datasets.

We are partnering with Howard University to build high quality African-American English (AAE) datasets to improve our products and make them work well for more people. Our research on globally inclusive cultural representation and our publication of the Monk Skin Tone scale furthers our commitments to equitable representation of all people. The insights we gain and techniques we develop not only help us improve our own models, they also power large-scale studies of representation in popular media to inform and inspire more inclusive content creation around the world.

Monk Skin Tone (MST) Scale. See more at skintone.google.

With advances in generative image models, fair and inclusive representation of people remains a top priority. In the development pipeline, we are working to amplify underrepresented voices and to better integrate social context knowledge. We proactively address potential harms and bias using classifiers and filters, careful dataset analysis, and in-model mitigations such as fine-tuning, reasoning, few-shot prompting, data augmentation and controlled decoding, and our research showed that generative AI enables higher quality safety classifiers to be developed with far less data. We also released a powerful way to better tune models with less data giving developers more control of responsibility challenges in generative AI.

We have developed new state-of-the-art explainability methods to identify the role of training data on model behaviors. By combining training data attribution methods with agile classifiers, we found that we can identify mislabelled training examples. This makes it possible to reduce the noise in training data, leading to significant improvements in model accuracy.

We initiated several efforts to improve safety and transparency about online content. For example, we introduced SynthID, a tool for watermarking and identifying AI-generated images. SynthID is imperceptible to the human eye, doesn't compromise image quality, and allows the watermark to remain detectable, even after modifications like adding filters, changing colors, and saving with various lossy compression schemes.

We also launched About This Image to help people assess the credibility of images, showing information like an image's history, how it's used on other pages, and available metadata about an image. And we explored safety methods that have been developed in other fields, learning from established situations where there is low-risk tolerance.

SynthID generates an imperceptible digital watermark for AI-generated images.

Privacy remains an essential aspect of our commitment to Responsible AI. We continued improving our state-of-the-art privacy preserving learning algorithm DP-FTRL, developed the DP-Alternating Minimization algorithm (DP-AM) to enable personalized recommendations with rigorous privacy protection, and defined a new general paradigm to reduce the privacy costs for many aggregation and learning tasks. We also proposed a scheme for auditing differentially private machine learning systems.

On the applications front we demonstrated that DP-SGD offers a practical solution in the large model fine-tuning regime and showed that images generated by DP diffusion models are useful for a range of downstream tasks. We proposed a new algorithm for DP training of large embedding models that provides efficient training on TPUs without compromising accuracy.

We also teamed up with a broad group of academic and industrial researchers to organize the first Machine Unlearning Challenge to address the scenario in which training images are forgotten to protect the privacy or rights of individuals. We shared a mechanism for extractable memorization, and participatory systems that give users more control over their sensitive data.

We continued to expand the world’s largest corpus of atypical speech recordings to >1M utterances in Project Euphonia, which enabled us to train a Universal Speech Model to better recognize atypical speech by 37% on real-world benchmarks.

We also built an audiobook recommendation system for students with reading disabilities such as dyslexia.


Adversarial testing

Our work in adversarial testing engaged community voices from historically marginalized communities. We partnered with groups such as the Equitable AI Research Round Table (EARR) to ensure we represent the diverse communities who use our models and engage with external users to identify potential harms in generative model outputs.

We established a dedicated Google AI Red Team focused on testing AI models and products for security, privacy, and abuse risks. We showed that attacks such as “poisoning” or adversarial examples can be applied to production models and surface additional risks such as memorization in both image and text generative models. We also demonstrated that defending against such attacks can be challenging, as merely applying defenses can cause other security and privacy leakages. We also introduced model evaluation for extreme risks, such as offensive cyber capabilities or strong manipulation skills.


Democratizing AI though tools and education

As we advance the state-of-the-art in ML and AI, we also want to ensure people can understand and apply AI to specific problems. We released MakerSuite (now Google AI Studio), a web-based tool that enables AI developers to quickly iterate and build lightweight AI-powered apps. To help AI engineers better understand and debug AI, we released LIT 1.0, a state-of-the-art, open-source debugger for machine learning models.

Colab, our tool that helps developers and students access powerful computing resources right in their web browser, reached over 10 million users. We’ve just added AI-powered code assistance to all users at no cost — making Colab an even more helpful and integrated experience in data and ML workflows.

One of the most used features is “Explain error” — whenever the user encounters an execution error in Colab, the code assistance model provides an explanation along with a potential fix.

To ensure AI produces accurate knowledge when put to use, we also recently introduced FunSearch, a new approach that generates verifiably true knowledge in mathematical sciences using evolutionary methods and large language models.

For AI engineers and product designers, we’re updating the People + AI Guidebook with generative AI best practices, and we continue to design AI Explorables, which includes how and why models sometimes make incorrect predictions confidently.


Community engagement

We continue to advance the fields of AI and computer science by publishing much of our work and participating in and organizing conferences. We have published more than 500 papers so far this year, and have strong presences at conferences like ICML (see the Google Research and Google DeepMind posts), ICLR (Google Research, Google DeepMind), NeurIPS (Google Research, Google DeepMind), ICCV, CVPR, ACL, CHI, and Interspeech. We are also working to support researchers around the world, participating in events like the Deep Learning Indaba, Khipu, supporting PhD Fellowships in Latin America, and more. We also worked with partners from 33 academic labs to pool data from 22 different robot types and create the Open X-Embodiment dataset and RT-X model to better advance responsible AI development.

Google has spearheaded an industry-wide effort to develop AI safety benchmarks under the MLCommons standards organization with participation from several major players in the generative AI space including OpenAI, Anthropic, Microsoft, Meta, Hugging Face, and more. Along with others in the industry we also co-founded the Frontier Model Forum (FMF), which is focused on ensuring safe and responsible development of frontier AI models. With our FMF partners and other philanthropic organizations, we launched a $10 million AI Safety Fund to advance research into the ongoing development of the tools for society to effectively test and evaluate the most capable AI models.

In close partnership with Google.org, we worked with the United Nations to build the UN Data Commons for the Sustainable Development Goals, a tool that tracks metrics across the 17 Sustainable Development Goals, and supported projects from NGOs, academic institutions, and social enterprises on using AI to accelerate progress on the SDGs.

The items highlighted in this post are a small fraction of the research work we have done throughout the last year. Find out more at the Google Research and Google DeepMind blogs, and our list of publications.


Future vision

As multimodal models become even more capable, they will empower people to make incredible progress in areas from science to education to entirely new areas of knowledge.

Progress continues apace, and as the year advances, and our products and research advance as well, people will find more and interesting creative uses for AI.

Ending this Year-in-Review where we began, as we say in Why We Focus on AI (and to what end):

If pursued boldly and responsibly, we believe that AI can be a foundational technology that transforms the lives of people everywhere — this is what excites us!

This Year-in-Review is cross-posted on both the Google Research Blog and the Google DeepMind Blog.

Source: Google AI Blog


VideoPoet: A large language model for zero-shot video generation

Posted by Dan Kondratyuk and David Ross, Software Engineers, Google Research

A recent wave of video generation models has burst onto the scene, in many cases showcasing stunning picturesque quality. One of the current bottlenecks in video generation is in the ability to produce coherent large motions. In many cases, even the current leading models either generate small motion or, when producing larger motions, exhibit noticeable artifacts.

To explore the application of language models in video generation, we introduce VideoPoet, a large language model (LLM) that is capable of a wide variety of video generation tasks, including text-to-video, image-to-video, video stylization, video inpainting and outpainting, and video-to-audio. One notable observation is that the leading video generation models are almost exclusively diffusion-based (for one example, see Imagen Video). On the other hand, LLMs are widely recognized as the de facto standard due to their exceptional learning capabilities across various modalities, including language, code, and audio (e.g., AudioPaLM). In contrast to alternative models in this space, our approach seamlessly integrates many video generation capabilities within a single LLM, rather than relying on separately trained components that specialize on each task.


Overview

The diagram below illustrates VideoPoet’s capabilities. Input images can be animated to produce motion, and (optionally cropped or masked) video can be edited for inpainting or outpainting. For stylization, the model takes in a video representing the depth and optical flow, which represent the motion, and paints contents on top to produce the text-guided style.

An overview of VideoPoet, capable of multitasking on a variety of video-centric inputs and outputs. The LLM can optionally take text as input to guide generation for text-to-video, image-to-video, video-to-audio, stylization, and outpainting tasks. Resources used: Wikimedia Commons and DAVIS.

Language models as video generators

One key advantage of using LLMs for training is that one can reuse many of the scalable efficiency improvements that have been introduced in existing LLM training infrastructure. However, LLMs operate on discrete tokens, which can make video generation challenging. Fortunately, there exist video and audio tokenizers, which serve to encode video and audio clips as sequences of discrete tokens (i.e., integer indices), and which can also be converted back into the original representation.

VideoPoet trains an autoregressive language model to learn across video, image, audio, and text modalities through the use of multiple tokenizers (MAGVIT V2 for video and image and SoundStream for audio). Once the model generates tokens conditioned on some context, these can be converted back into a viewable representation with the tokenizer decoders.

A detailed look at the VideoPoet task design, showing the training and inference inputs and outputs of various tasks. Modalities are converted to and from tokens using tokenizer encoder and decoders. Each modality is surrounded by boundary tokens, and a task token indicates the type of task to perform.

Examples generated by VideoPoet

Some examples generated by our model are shown below.

Videos generated by VideoPoet from various text prompts. For specific text prompts refer to the website.

For text-to-video, video outputs are variable length and can apply a range of motions and styles depending on the text content. To ensure responsible practices, we reference artworks and styles in the public domain e.g., Van Gogh’s “Starry Night”.

Text Input    “A Raccoon dancing in Times Square”    “A horse galloping through Van-Gogh’s ‘Starry Night’”    “Two pandas playing cards”    “A large blob of exploding splashing rainbow paint, with an apple emerging, 8k”
Video Output            

For image-to-video, VideoPoet can take the input image and animate it with a prompt.

An example of image-to-video with text prompts to guide the motion. Each video is paired with an image to its left. Left: “A ship navigating the rough seas, thunderstorm and lightning, animated oil on canvas”. Middle: “Flying through a nebula with many twinkling stars”. Right: “A wanderer on a cliff with a cane looking down at the swirling sea fog below on a windy day”. Reference: Wikimedia Commons, public domain**.

For video stylization, we predict the optical flow and depth information before feeding into VideoPoet with some additional input text.

Examples of video stylization on top of VideoPoet text-to-video generated videos with text prompts, depth, and optical flow used as conditioning. The left video in each pair is the input video, the right is the stylized output. Left: “Wombat wearing sunglasses holding a beach ball on a sunny beach.” Middle: “Teddy bears ice skating on a crystal clear frozen lake.” Right: “A metal lion roaring in the light of a forge.”

VideoPoet is also capable of generating audio. Here we first generate 2-second clips from the model and then try to predict the audio without any text guidance. This enables generation of video and audio from a single model.



        

An example of video-to-audio, generating audio from a video example without any text input.

By default, the VideoPoet model generates videos in portrait orientation to tailor its output towards short-form content. To showcase its capabilities, we have produced a brief movie composed of many short clips generated by VideoPoet. For the script, we asked Bard to write a short story about a traveling raccoon with a scene-by-scene breakdown and a list of accompanying prompts. We then generated video clips for each prompt, and stitched together all resulting clips to produce the final video below.




When we developed VideoPoet, we noticed some nice properties of the model’s capabilities, which we highlight below.


Long video

We are able to generate longer videos simply by conditioning on the last 1 second of video and predicting the next 1 second. By chaining this repeatedly, we show that the model can not only extend the video well but also faithfully preserve the appearance of all objects even over several iterations.

Here are two examples of VideoPoet generating long video from text input:

Text Input    “An astronaut starts dancing on Mars. Colorful fireworks then explode in the background.”    “FPV footage of a very sharp elven city of stone in the jungle with a brilliant blue river, waterfall, and large steep vertical cliff faces.”           
Video Output                 

It is also possible to interactively edit existing video clips generated by VideoPoet. If we supply an input video, we can change the motion of objects to perform different actions. The object manipulation can be centered at the first frame or the middle frames, which allow for a high degree of editing control.

For example, we can randomly generate some clips from the input video and select the desired next clip.

An input video on the left is used as conditioning to generate four choices given the initial prompt: “Closeup of an adorable rusty broken-down steampunk robot covered in moss moist and budding vegetation, surrounded by tall grass”. For the first three outputs we show what would happen for unprompted motions. For the last video in the list below, we add to the prompt, “powering up with smoke in the background” to guide the action.

Image to video control

Similarly, we can apply motion to an input image to edit its contents towards the desired state, conditioned on a text prompt.

Animating a painting with different prompts. Left: “A woman turning to look at the camera.” Right: “A woman yawning.” **

Camera motion

We can also accurately control camera movements by appending the type of desired camera motion to the text prompt. As an example, we generated an image by our model with the prompt, “Adventure game concept art of a sunrise over a snowy mountain by a crystal clear river”. The examples below append the given text suffix to apply the desired motion.

Prompts from left to right: “Zoom out”, “Dolly zoom”, “Pan left”, “Arc shot”, “Crane shot”, “FPV drone shot”.

Evaluation results

We evaluate VideoPoet on text-to-video generation with a variety of benchmarks to compare the results to other approaches. To ensure a neutral evaluation, we ran all models on a wide variation of prompts without cherry-picking examples and asked people to rate their preferences. The figure below highlights the percentage of the time VideoPoet was chosen as the preferred option in green for the following questions.


Text fidelity

User preference ratings for text fidelity, i.e., what percentage of videos are preferred in terms of accurately following a prompt.

Motion interestingness

User preference ratings for motion interestingness, i.e., what percentage of videos are preferred in terms of producing interesting motion.

Based on the above, on average people selected 24–35% of examples from VideoPoet as following prompts better than a competing model vs. 8–11% for competing models. Raters also preferred 41–54% of examples from VideoPoet for more interesting motion than 11–21% for other models.


Conclusion

Through VideoPoet, we have demonstrated LLMs’ highly-competitive video generation quality across a wide variety of tasks, especially in producing interesting and high quality motions within videos. Our results suggest the promising potential of LLMs in the field of video generation. For future directions, our framework should be able to support “any-to-any” generation, e.g., extending to text-to-audio, audio-to-video, and video captioning should be possible, among many others.

To view more examples in original quality, see the website demo.


Acknowledgements

This research has been supported by a large body of contributors, including Dan Kondratyuk, Lijun Yu, Xiuye Gu, José Lezama, Jonathan Huang, Rachel Hornung, Hartwig Adam, Hassan Akbari, Yair Alon, Vighnesh Birodkar, Yong Cheng, Ming-Chang Chiu, Josh Dillon, Irfan Essa, Agrim Gupta, Meera Hahn, Anja Hauth, David Hendon, Alonso Martinez, David Minnen, David Ross, Grant Schindler, Mikhail Sirotenko, Kihyuk Sohn, Krishna Somandepalli, Huisheng Wang, Jimmy Yan, Ming-Hsuan Yang, Xuan Yang, Bryan Seybold, and Lu Jiang.

We give special thanks to Alex Siegman and Victor Gomes for managing computing resources. We also give thanks to Aren Jansen, Marco Tagliasacchi, Neil Zeghidour, John Hershey for audio tokenization and processing, Angad Singh for storyboarding in “Rookie the Raccoon”, Cordelia Schmid for research discussions, Alonso Martinez for graphic design, David Salesin, Tomas Izo, and Rahul Sukthankar for their support, and Jay Yagnik as architect of the initial concept.


**
(a) The Storm on the Sea of Galilee, by Rembrandt 1633, public domain.
(b) Pillars of Creation, by NASA 2014, public domain.
(c) Wanderer above the Sea of Fog, by Caspar David Friedrich, 1818, public domain
(d) Mona Lisa, by Leonardo Da Vinci, 1503, public domain.

Source: Google AI Blog


Simulations illuminate the path to post-event traffic flow

Posted by Yechen Li and Neha Arora, Software Engineers, Google Research

Fifteen minutes. That’s how long it took to empty the Colosseum, an engineering marvel that’s still standing as the largest amphitheater in the world. Two thousand years later, this design continues to work well to move enormous crowds out of sporting and entertainment venues.

But of course, exiting the arena is only the first step. Next, people must navigate the traffic that builds up in the surrounding streets. This is an age-old problem that remains unsolved to this day. In Rome, they addressed the issue by prohibiting private traffic on the street that passes directly by the Colosseum. This policy worked there, but what if you’re not in Rome? What if you’re at the Superbowl? Or at a Taylor Swift concert?

An approach to addressing this problem is to use simulation models, sometimes called "digital twins", which are virtual replicas of real-world transportation networks that attempt to capture every detail from the layout of streets and intersections to the flow of vehicles. These models allow traffic experts to mitigate congestion, reduce accidents, and improve the experience of drivers, riders, and walkers alike. Previously, our team used these models to quantify sustainability impact of routing, test evacuation plans and show simulated traffic in Maps Immersive View.

Calibrating high-resolution traffic simulations to match the specific dynamics of a particular setting is a longstanding challenge in the field. The availability of aggregate mobility data, detailed Google Maps road network data, advances in transportation science (such as understanding the relationship between segment demands and speeds for road segments with traffic signals), and calibration techniques which make use of speed data in physics-informed traffic models are paving the way for compute-efficient optimization at a global scale.

To test this technology in the real world, Google Research partnered with the Seattle Department of Transportation (SDOT) to develop simulation-based traffic guidance plans. Our goal is to help thousands of attendees of major sports and entertainment events leave the stadium area quickly and safely. The proposed plan reduced average trip travel times by 7 minutes for vehicles leaving the stadium region during large events. We deployed it in collaboration with SDOT using Dynamic Message Signs (DMS) and verified impact over multiple events between August and November, 2023.

One policy recommendation we made was to divert traffic from S Spokane St, a major thoroughfare that connects the area to highways I-5 and SR 99, and is often congested after events. Suggested changes improved the flow of traffic through highways and arterial streets near the stadium, and reduced the length of vehicle queues that formed behind traffic signals. (Note that vehicles are larger than reality in this clip for demonstration.)

Simulation model

For this project, we created a new simulation model of the area around Seattle’s stadiums. The intent for this model is to replay each traffic situation for a specified day as closely as possible. We use an open-source simulation software, Simulation of Urban MObility (SUMO). SUMO’s behavioral models help us describe traffic dynamics, for instance, how drivers make decisions, like car-following, lane-changing and speed limit compliance. We also use insights from Google Maps to define the network’s structure and various static segment attributes (e.g., number of lanes, speed limit, presence of traffic lights).

Overview of the Simulation framework.

Travel demand is an important simulator input. To compute it, we first decompose the road network of a given metropolitan area into zones, specifically level 13 S2 cells with 1.27 km2 area per cell. From there, we define the travel demand as the expected number of trips that travel from an origin zone to a destination zone in a given time period. The demand is represented as aggregated origin–destination (OD) matrices.

To get the initial expected number of trips between an origin zone and a destination zone, we use aggregated and anonymized mobility statistics. Then we solve the OD calibration problem by combining initial demand with observed traffic statistics, like segment speeds, travel times and vehicular counts, to reproduce event scenarios.

We model the traffic around multiple past events in Seattle’s T-Mobile Park and Lumen Field and evaluate the accuracy by computing aggregated and anonymized traffic statistics. Analyzing these event scenarios helps us understand the effect of different routing policies on congestion in the region.

Heatmaps demonstrate a substantial increase in numbers of trips in the region after a game as compared to the same time on a non-game day.
The graph shows observed segment speeds on the x-axis and simulated speeds on the y-axis for a modeled event. The concentration of data points along the red x=y line demonstrates the ability of the simulation to reproduce realistic traffic conditions.

Routing policies

SDOT and the Seattle Police Department’s (SPD) local knowledge helped us determine the most congested routes that needed improvement:

  • Traffic from T-Mobile Park stadium parking lot’s Edgar Martinez Dr. S exit to eastbound I-5 highway / westbound SR 99 highway
  • Traffic through Lumen Field stadium parking lot to northbound Cherry St. I-5 on-ramp
  • Traffic going southbound through Seattle’s SODO neighborhood to S Spokane St.

We developed routing policies and evaluated them using the simulation model. To disperse traffic faster, we tried policies that would route northbound/southbound traffic from the nearest ramps to further highway ramps, to shorten the wait times. We also experimented with opening HOV lanes to event traffic, recommending alternate routes (e.g., SR 99), or load sharing between different lanes to get to the nearest stadium ramps.


Evaluation results

We model multiple events with different traffic conditions, event times, and attendee counts. For each policy, the simulation reproduces post-game traffic and reports the travel time for vehicles, from departing the stadium to reaching their destination or leaving the Seattle SODO area. The time savings are computed as the difference of travel time before/after the policy, and are shown in the below table, per policy, for small and large events. We apply each policy to a percentage of traffic, and re-estimate the travel times. Results are shown if 10%, 30%, or 50% of vehicles are affected by a policy.

Based on these simulation results, the feasibility of implementation, and other considerations, SDOT has decided to implement the “Northbound Cherry St ramp” and “Southbound S Spokane St ramp” policies using DMS during large events. The signs suggest drivers take alternative routes to reach their destinations. The combination of these two policies leads to an average of 7 minutes of travel time savings per vehicle, based on rerouting 30% of traffic during large events.


Conclusion

This work demonstrates the power of simulations to model, identify, and quantify the effect of proposed traffic guidance policies. Simulations allow network planners to identify underused segments and evaluate the effects of different routing policies, leading to a better spatial distribution of traffic. The offline modeling and online testing show that our approach can reduce total travel time. Further improvements can be made by adding more traffic management strategies, such as optimizing traffic lights. Simulation models have been historically time consuming and hence affordable only for the largest cities and high stake projects. By investing in more scalable techniques, we hope to bring these models to more cities and use cases around the world.


Acknowledgements

In collaboration with Alex Shashko, Andrew Tomkins, Ashley Carrick, Carolina Osorio, Chao Zhang, Damien Pierce, Iveel Tsogsuren, Sheila de Guia, and Yi-fan Chen. Visual design by John Guilyard. We would like to thank our SDOT partners Carter Danne, Chun Kwan, Ethan Bancroft, Jason Cambridge, Laura Wojcicki, Michael Minor, Mohammed Said, Trevor Partap, and SPD partners Lt. Bryan Clenna and Sgt. Brian Kokesh.

Source: Google AI Blog


Advancements in machine learning for machine learning

Posted by Phitchaya Mangpo Phothilimthana, Staff Research Scientist, Google DeepMind, and Bryan Perozzi, Senior Staff Research Scientist, Google Research

With the recent and accelerated advances in machine learning (ML), machines can understand natural language, engage in conversations, draw images, create videos and more. Modern ML models are programmed and trained using ML programming frameworks, such as TensorFlow, JAX, PyTorch, among many others. These libraries provide high-level instructions to ML practitioners, such as linear algebra operations (e.g., matrix multiplication, convolution, etc.) and neural network layers (e.g., 2D convolution layers, transformer layers). Importantly, practitioners need not worry about how to make their models run efficiently on hardware because an ML framework will automatically optimize the user's model through an underlying compiler. The efficiency of the ML workload, thus, depends on how good the compiler is. A compiler typically relies on heuristics to solve complex optimization problems, often resulting in suboptimal performance.

In this blog post, we present exciting advancements in ML for ML. In particular, we show how we use ML to improve efficiency of ML workloads! Prior works, both internal and external, have shown that we can use ML to improve performance of ML programs by selecting better ML compiler decisions. Although there exist a few datasets for program performance prediction, they target small sub-programs, such as basic blocks or kernels. We introduce “TpuGraphs: A Performance Prediction Dataset on Large Tensor Computational Graphs” (presented at NeurIPS 2023), which we recently released to fuel more research in ML for program optimization. We hosted a Kaggle competition on the dataset, which recently completed with 792 participants on 616 teams from 66 countries. Furthermore, in “Learning Large Graph Property Prediction via Graph Segment Training”, we cover a novel method to scale graph neural network (GNN) training to handle large programs represented as graphs. The technique both enables training arbitrarily large graphs on a device with limited memory capacity and improves generalization of the model.


ML compilers

ML compilers are software routines that convert user-written programs (here, mathematical instructions provided by libraries such as TensorFlow) to executables (instructions to execute on the actual hardware). An ML program can be represented as a computation graph, where a node represents a tensor operation (such as matrix multiplication), and an edge represents a tensor flowing from one node to another. ML compilers have to solve many complex optimization problems, including graph-level and kernel-level optimizations. A graph-level optimization requires the context of the entire graph to make optimal decisions and transforms the entire graph accordingly. A kernel-level optimization transforms one kernel (a fused subgraph) at a time, independently of other kernels.

Important optimizations in ML compilers include graph-level and kernel-level optimizations.

To provide a concrete example, imagine a matrix (2D tensor):

It can be stored in computer memory as [A B C a b c] or [A a B b C c], known as row- and column-major memory layout, respectively. One important ML compiler optimization is to assign memory layouts to all intermediate tensors in the program. The figure below shows two different layout configurations for the same program. Let’s assume that on the left-hand side, the assigned layouts (in red) are the most efficient option for each individual operator. However, this layout configuration requires the compiler to insert a copy operation to transform the memory layout between the add and convolution operations. On the other hand, the right-hand side configuration might be less efficient for each individual operator, but it doesn’t require the additional memory transformation. The layout assignment optimization has to trade off between local computation efficiency and layout transformation overhead.

A node represents a tensor operator, annotated with its output tensor shape [n0, n1, ...], where ni is the size of dimension i. Layout {d0, d1, ...} represents minor-to-major ordering in memory. Applied configurations are highlighted in red, and other valid configurations are highlighted in blue. A layout configuration specifies the layouts of inputs and outputs of influential operators (i.e., convolution and reshape). A copy operator is inserted when there is a layout mismatch.

If the compiler makes optimal choices, significant speedups can be made. For example, we have seen up to a 32% speedup when choosing an optimal layout configuration over the default compiler’s configuration in the XLA benchmark suite.


TpuGraphs dataset

Given the above, we aim to improve ML model efficiency by improving the ML compiler. Specifically, it can be very effective to equip the compiler with a learned cost model that takes in an input program and compiler configuration and then outputs the predicted runtime of the program.

With this motivation, we release TpuGraphs, a dataset for learning cost models for programs running on Google’s custom Tensor Processing Units (TPUs). The dataset targets two XLA compiler configurations: layout (generalization of row- and column-major ordering, from matrices, to higher dimension tensors) and tiling (configurations of tile sizes). We provide download instructions and starter code on the TpuGraphs GitHub. Each example in the dataset contains a computational graph of an ML workload, a compilation configuration, and the execution time of the graph when compiled with the configuration. The graphs in the dataset are collected from open-source ML programs, featuring popular model architectures, e.g., ResNet, EfficientNet, Mask R-CNN, and Transformer. The dataset provides 25× more graphs than the largest (earlier) graph property prediction dataset (with comparable graph sizes), and graph size is 770× larger on average compared to existing performance prediction datasets on ML programs. With this greatly expanded scale, for the first time we can explore the graph-level prediction task on large graphs, which is subject to challenges such as scalability, training efficiency, and model quality.

Scale of TpuGraphs compared to other graph property prediction datasets.

We provide baseline learned cost models with our dataset (architecture shown below). Our baseline models are based on a GNN since the input program is represented as a graph. Node features, shown in blue below, consist of two parts. The first part is an opcode id, the most important information of a node, which indicates the type of tensor operation. Our baseline models, thus, map an opcode id to an opcode embedding via an embedding lookup table. The opcode embedding is then concatenated with the second part, the rest of the node features, as inputs to a GNN. We combine the node embeddings produced by the GNN to create the fixed-size embedding of the graph using a simple graph pooling reduction (i.e., sum and mean). The resulting graph embedding is then linearly transformed into the final scalar output by a feedforward layer.

Our baseline learned cost model employs a GNN since programs can be naturally represented as graphs.

Furthermore we present Graph Segment Training (GST), a method for scaling GNN training to handle large graphs on a device with limited memory capacity in cases where the prediction task is on the entire-graph (i.e., graph-level prediction). Unlike scaling training for node- or edge-level prediction, scaling for graph-level prediction is understudied but crucial to our domain, as computation graphs can contain hundreds of thousands of nodes. In a typical GNN training (“Full Graph Training”, on the left below), a GNN model is trained using an entire graph, meaning all nodes and edges of the graph are used to compute gradients. For large graphs, this might be computationally infeasible. In GST, each large graph is partitioned into smaller segments, and a random subset of segments is selected to update the model; embeddings for the remaining segments are produced without saving their intermediate activations (to avoid consuming memory). The embeddings of all segments are then combined to generate an embedding for the original large graph, which is then used for prediction. In addition, we introduce the historical embedding table to efficiently obtain graph segments’ embeddings and segment dropout to mitigate the staleness from historical embeddings. Together, our complete method speeds up the end-to-end training time by 3×.

Comparing Full Graph Training (typical method) vs Graph Segment Training (our proposed method).

Kaggle competition

Finally, we ran the “Fast or Slow? Predict AI Model Runtime” competition over the TpuGraph dataset. This competition ended with 792 participants on 616 teams. We had 10507 submissions from 66 countries. For 153 users (including 47 in the top 100), this was their first competition. We learned many interesting new techniques employed by the participating teams, such as:

  • Graph pruning / compression: Instead of using the GST method, many teams experimented with different ways to compress large graphs (e.g., keeping only subgraphs that include the configurable nodes and their immediate neighbors).
  • Feature padding value: Some teams observed that the default padding value of 0 is problematic because 0 clashes with a valid feature value, so using a padding value of -1 can improve the model accuracy significantly.
  • Node features: Some teams observed that additional node features (such as dot general’s contracting dimensions) are important. A few teams found that different encodings of node features also matter.
  • Cross-configuration attention: A winning team designed a simple layer that allows the model to explicitly "compare" configs against each other. This technique is shown to be much better than letting the model infer for each config individually.

We will debrief the competition and preview the winning solutions at the competition session at the ML for Systems workshop at NeurIPS on December 16, 2023. Finally, congratulations to all the winners and thank you for your contributions to advancing research in ML for systems!


NeurIPS expo

If you are interested in more research about structured data and artificial intelligence, we hosted the NeurIPS Expo panel Graph Learning Meets Artificial Intelligence on December 9, which covered advancing learned cost models and more!


Acknowledgements

Sami Abu-el-Haija (Google Research) contributed significantly to this work and write-up. The research in this post describes joint work with many additional collaborators including Mike Burrows, Kaidi Cao, Bahare Fatemi, Jure Leskovec, Charith Mendis, Dustin Zelle, and Yanqi Zhou.

Source: Google AI Blog


StyleDrop: Text-to-image generation in any style

Posted by Kihyuk Sohn and Dilip Krishnan, Research Scientists, Google Research

Text-to-image models trained on large volumes of image-text pairs have enabled the creation of rich and diverse images encompassing many genres and themes. Moreover, popular styles such as “anime” or “steampunk”, when added to the input text prompt, may translate to specific visual outputs. While many efforts have been put into prompt engineering, a wide range of styles are simply hard to describe in text form due to the nuances of color schemes, illumination, and other characteristics. As an example, “watercolor painting” may refer to various styles, and using a text prompt that simply says “watercolor painting style” may either result in one specific style or an unpredictable mix of several.

When we refer to "watercolor painting style," which do we mean? Instead of specifying the style in natural language, StyleDrop allows the generation of images that are consistent in style by referring to a style reference image*.

In this blog we introduce “StyleDrop: Text-to-Image Generation in Any Style”, a tool that allows a significantly higher level of stylized text-to-image synthesis. Instead of seeking text prompts to describe the style, StyleDrop uses one or more style reference images that describe the style for text-to-image generation. By doing so, StyleDrop enables the generation of images in a style consistent with the reference, while effectively circumventing the burden of text prompt engineering. This is done by efficiently fine-tuning the pre-trained text-to-image generation models via adapter tuning on a few style reference images. Moreover, by iteratively fine-tuning the StyleDrop on a set of images it generated, it achieves the style-consistent image generation from text prompts.


Method overview

StyleDrop is a text-to-image generation model that allows generation of images whose visual styles are consistent with the user-provided style reference images. This is achieved by a couple of iterations of parameter-efficient fine-tuning of pre-trained text-to-image generation models. Specifically, we build StyleDrop on Muse, a text-to-image generative vision transformer.


Muse: text-to-image generative vision transformer

Muse is a state-of-the-art text-to-image generation model based on the masked generative image transformer (MaskGIT). Unlike diffusion models, such as Imagen or Stable Diffusion, Muse represents an image as a sequence of discrete tokens and models their distribution using a transformer architecture. Compared to diffusion models, Muse is known to be faster while achieving competitive generation quality.


Parameter-efficient adapter tuning

StyleDrop is built by fine-tuning the pre-trained Muse model on a few style reference images and their corresponding text prompts. There have been many works on parameter-efficient fine-tuning of transformers, including prompt tuning and Low-Rank Adaptation (LoRA) of large language models. Among those, we opt for adapter tuning, which is shown to be effective at fine-tuning a large transformer network for language and image generation tasks in a parameter-efficient manner. For example, it introduces less than one million trainable parameters to fine-tune a Muse model of 3B parameters, and it requires only 1000 training steps to converge.

Parameter-efficient adapter tuning of Muse.

Iterative training with feedback

While StyleDrop is effective at learning styles from a few style reference images, it is still challenging to learn from a single style reference image. This is because the model may not effectively disentangle the content (i.e., what is in the image) and the style (i.e., how it is being presented), leading to reduced text controllability in generation. For example, as shown below in Step 1 and 2, a generated image of a chihuahua from StyleDrop trained from a single style reference image shows a leakage of content (i.e., the house) from the style reference image. Furthermore, a generated image of a temple looks too similar to the house in the reference image (concept collapse).

We address this issue by training a new StyleDrop model on a subset of synthetic images, chosen by the user or by image-text alignment models (e.g., CLIP), whose images are generated by the first round of the StyleDrop model trained on a single image. By training on multiple synthetic image-text aligned images, the model can easily disentangle the style from the content, thus achieving improved image-text alignment.

Iterative training with feedback*. The first round of StyleDrop may result in reduced text controllability, such as a content leakage or concept collapse, due to the difficulty of content-style disentanglement. Iterative training using synthetic images, generated by the previous rounds of StyleDrop models and chosen by human or image-text alignment models, improves the text adherence of stylized text-to-image generation.

Experiments


StyleDrop gallery

We show the effectiveness of StyleDrop by running experiments on 24 distinct style reference images. As shown below, the images generated by StyleDrop are highly consistent in style with each other and with the style reference image, while depicting various contexts, such as a baby penguin, banana, piano, etc. Moreover, the model can render alphabet images with a consistent style.

Stylized text-to-image generation. Style reference images* are on the left inside the yellow box. Text prompts used are:
First row: a baby penguin, a banana, a bench.
Second row: a butterfly, an F1 race car, a Christmas tree.
Third row: a coffee maker, a hat, a moose.
Fourth row: a robot, a towel, a wood cabin.
Stylized visual character generation. Style reference images* are on the left inside the yellow box. Text prompts used are: (first row) letter 'A', letter 'B', letter 'C', (second row) letter 'E', letter 'F', letter 'G'.

Generating images of my object in my style

Below we show generated images by sampling from two personalized generation distributions, one for an object and another for the style.

Images at the top in the blue border are object reference images from the DreamBooth dataset (teapot, vase, dog and cat), and the image on the left at the bottom in the red border is the style reference image*. Images in the purple border (i.e. the four lower right images) are generated from the style image of the specific object.

Quantitative results

For the quantitative evaluation, we synthesize images from a subset of Parti prompts and measure the image-to-image CLIP score for style consistency and image-to-text CLIP score for text consistency. We study non–fine-tuned models of Muse and Imagen. Among fine-tuned models, we make a comparison to DreamBooth on Imagen, state-of-the-art personalized text-to-image method for subjects. We show two versions of StyleDrop, one trained from a single style reference image, and another, “StyleDrop (HF)”, that is trained iteratively using synthetic images with human feedback as described above. As shown below, StyleDrop (HF) shows significantly improved style consistency score over its non–fine-tuned counterpart (0.694 vs. 0.556), as well as DreamBooth on Imagen (0.694 vs. 0.644). We observe an improved text consistency score with StyleDrop (HF) over StyleDrop (0.322 vs. 0.313). In addition, in a human preference study between DreamBooth on Imagen and StyleDrop on Muse, we found that 86% of the human raters preferred StyleDrop on Muse over DreamBooth on Imagen in terms of consistency to the style reference image.


Conclusion

StyleDrop achieves style consistency at text-to-image generation using a few style reference images. Google’s AI Principles guided our development of Style Drop, and we urge the responsible use of the technology. StyleDrop was adapted to create a custom style model in Vertex AI, and we believe it could be a helpful tool for art directors and graphic designers — who might want to brainstorm or prototype visual assets in their own styles, to improve their productivity and boost their creativity — or businesses that want to generate new media assets that reflect a particular brand. As with other generative AI capabilities, we recommend that practitioners ensure they align with copyrights of any media assets they use. More results are found on our project website and YouTube video.


Acknowledgements

This research was conducted by Kihyuk Sohn, Nataniel Ruiz, Kimin Lee, Daniel Castro Chin, Irina Blok, Huiwen Chang, Jarred Barber, Lu Jiang, Glenn Entis, Yuanzhen Li, Yuan Hao, Irfan Essa, Michael Rubinstein, and Dilip Krishnan. We thank owners of images used in our experiments (links for attribution) for sharing their valuable assets.

*See image sources ↩

Source: Google AI Blog


Google at NeurIPS 2023

Posted by Catherine Armato, Program Manager, Google

This week the 37th annual Conference on Neural Information Processing Systems (NeurIPS 2023), the biggest machine learning conference of the year, kicks off in New Orleans, LA. Google is proud to be a Diamond Level sponsor of NeurIPS this year and will have a strong presence with >170 accepted papers, two keynote talks, and additional contributions to the broader research community through organizational support and involvement in >20 workshops and tutorials. Google is also proud to be a Platinum Sponsor for both the Women in Machine Learning and LatinX in AI workshops. We look forward to sharing some of our extensive ML research and expanding our partnership with the broader ML research community.

Attending for NeurIPS 2023 in person? Come visit the Google Research booth to learn more about the exciting work we’re doing to solve some of the field’s most interesting challenges. Visit the @GoogleAI X (Twitter) account to find out about Google booth activities (e.g., demos and Q&A sessions).

You can learn more about our latest cutting edge work being presented at the conference in the list below (Google affiliations highlighted in bold). And see Google DeepMind’s blog to learn more about their participation at NeurIPS 2023.


Board & Organizing Committee

NeurIPS Board: Corinna Cortes
Advisory Board: John C. Platt
Senior Area Chair: Inderjit S. Dhillon
Creative AI Chair: Isabelle Guyon
Program Chair: Amir Globerson
Datasets and Benchmarks Chair: Remi Denton


Google Research Booth Demo/Q&A Schedule

This schedule is subject to change. Please visit the Google booth (#215) for more information.

What You See is What You Read? Improving Text-Image Alignment Evaluation
Presenter: Yonatan Bitton
Monday, Dec 11 | 12:15PM - 1:45PM

Talk like a Graph: Encoding Graphs for Large Language Models
Presenters: Bahar Fatemi, Jonathan Halcrow, Bryan Perozzi
Monday, Dec 11 | 4:00PM - 4:45PM

VisIT-Bench: A Benchmark for Vision-Language Instruction Following Inspired by Real-World Use
Presenter: Yonatan Bitton
Monday, Dec 11 | 4:00PM - 4:45PM

MLCommons Croissant
Presenters: Omar Benjelloun, Meg Risdal, Lora Aroyo
Tuesday, Dec 12 | 9:15AM - 10:00AM

DaTaSeg: Taming a Universal Multi-Dataset Multi-Task Segmentation Model
Presenter: Xiuye Gu
Tuesday, Dec 12 | 12:45PM - 2:15PM

Embedding Large Graphs
Presenters: Bryan Perozzi, Anton Tsitsulin
Tuesday, Dec 12 | 3:20PM - 3:40PM

Correlated Noise Provably Beats Independent Noise for Differentially Private Learning
Presenter: Krishna Pillutla
Tuesday, Dec 12 | 3:20PM - 3:40PM

Med-PaLM
Presenter: Tao Tu
Tuesday, Dec 12 | 4:45PM - 5:15PM

StyleDrop: Text-to-Image Generation in Any Style
Presenters: Kihyuk Sohn, Lu Jiang, Irfan Essa
Tuesday, Dec 12 | 4:45PM - 5:15PM

DICES Dataset: Diversity in Conversational AI Evaluation for Safety
Presenters: Lora Aroyo, Alicia Parrish, Vinodkumar Prabhakaran
Wednesday, Dec 13 | 9:15AM - 10:00AM

Resonator: Scalable Game-Based Evaluation of Large Models
Presenters: Erin Drake Kajioka, Michal Todorovic
Wednesday, Dec 13 | 12:45PM - 2:15PM

Adversarial Nibbler
Presenter: Lora Aroyo
Wednesday, Dec 13 | 12:45PM - 2:15PM

Towards Generalist Biomedical AI
Presenter: Tao Tu
Wednesday, Dec 13 | 3:15PM - 3:30PM

Conditional Adaptors
Presenter: Junwen Bai
Wednesday, Dec 13 | 3:15PM - 3:30PM

Patient Assistance with Multimodal RAG
Presenters: Ryan Knuffman, Milica Cvetkovic
Wednesday, Dec 13 | 4:15PM - 5:00PM

How Hessian Structure Explains Mysteries in Sharpness Regularization
Presenter: Hossein Mobahi
Wednesday, Dec 13 | 4:15PM - 5:00PM


Keynote Speakers


Affinity Workshops

Women in ML
Google Sponsored - Platinum

LatinX in AI
Google Sponsored - Platinum

New in ML
Organizer: Isabelle Guyon


Workshops

AI for Accelerated Materials Design (AI4Mat-2023)
Fireside Chat: Gowoon Cheon

Associative Memory & Hopfield Networks in 2023
Panelist: Blaise Agüera y Arcas

Information-Theoretic Principles in Cognitive Systems (InfoCog)
Speaker: Alexander Alemi

Machine Learning and the Physical Sciences
Speaker: Alexander Alemi

UniReps: Unifying Representations in Neural Models
Organizer: Mathilde Caron

Robustness of Zero/Few-shot Learning in Foundation Models (R0-FoMo)
Speaker: Partha Talukdar
Organizer: Ananth Balashankar, Yao Qin, Ahmad Beirami

Workshop on Diffusion Models
Speaker: Tali Dekel

Algorithmic Fairness through the Lens of Time
Roundtable Lead: Stephen Pfohl
Organizer: Golnoosh Farnadi

Backdoors in Deep Learning: The Good, the Bad, and the Ugly
Organizer: Eugene Bagdasaryan

OPT 2023: Optimization for Machine Learning
Organizer: Cristóbal Guzmán

Machine Learning for Creativity and Design
Speaker: Aleksander Holynski, Alexander Mordvintsev

Robot Learning Workshop: Pretraining, Fine-Tuning, and Generalization with Large Scale Models
Speaker: Matt Barnes

Machine Learning for Audio
Organizer: Shrikanth Narayanan

Federated Learning in the Age of Foundation Models (FL@FM-NeurIPS’23)
Speaker: Cho-Jui Hsieh, Zheng Xu

Socially Responsible Language Modelling Research (SoLaR)
Panelist: Vinodkumar Prabhakaran

I Can’t Believe It’s Not Better (ICBINB): Failure Modes in the Age of Foundation Models
Advisory Board: Javier Antorán

Machine Learning for Systems
Organizer: Yawen Wang
Competition Committee: Bryan Perozzi, Sami Abu-el-haija
Steering Committee: Milad Hashemi

Self-Supervised Learning: Theory and Practice
Organizer: Mathilde Caron


Competitions

NeurIPS 2023 Machine Unlearning Competition
Organizer: Isabelle Guyon, Peter Kairouz

Lux AI Challenge Season 2 NeurIPS Edition
Organizer: Bovard Doerschuk-Tiberi, Addison Howard


Tutorials

Data-Centric AI for Reliable and Responsible AI: From Theory to Practice
Isabelle Guyon, Nabeel Seedat, Mihaela va der Schaar


Creative AI Track

Creative AI Performances 1 & 2
Speaker: Erin Drake Kajioka, Yonatan Bitton
Organizer: Isabelle Guyon
Performance 1: Mon, Dec 11 | 6:30PM - 8:30PM, Lobby Stage
Performance 2: Thu, Dec 14 | 7:00PM - 9:00PM, Lobby Stage

Creative AI Sessions 1 – 3
Speaker: Erin Drake Kajioka, Yonatan Bitton
Organizer: Isabelle Guyon
Session 1: Tue, Dec 12 | 3:05PM - 3:40PM, Hall D2
Session 2: Wed, Dec 13 | 10:45AM - 2:15PM, Hall D2
Session 3: Thu, Dec 14 | 10:45 AM - 2:15PM, Hall D2

Creative AI Videos
Organizer: Isabelle Guyon


Expo Talks

Graph Learning Meets Artificial Intelligence
Speaker: Bryan Perozzi

Resonator: Music Space
Speakers: Erin Drake Kajioka, Michal Todorovic

Empirical Rigor in ML as a Massively Parallelizable Challenge
Speaker: Megan Risdal (Kaggle)


Oral Talks

Ordering-based Conditions for Global Convergence of Policy Gradient Methods
Jincheng Mei, Bo Dai, Alekh Agarwal, Mohammad Ghavamzadeh*, Csaba Szepesvari, Dale Schuurmans

Private Everlasting Prediction
Moni Naor, Kobbi Nissim, Uri Stemmer, Chao Yan

User-Level Differential Privacy With Few Examples Per User
Badih Ghazi, Pritish Kamath, Ravi Kumar, Pasin Manurangsi, Raghu Meka, Chiyuan Zhang

DataComp: In Search of the Next Generation of Multimodal Datasets
Samir Yitzhak Gadre, Gabriel Ilharco, Alex Fang, Jonathan Hayase, Georgios Smyrnis, Thao Nguyen, Ryan Marten, Mitchell Wortsman, Dhruba Ghosh, Jieyu Zhang, Eyal Orgad, Rahim Entezari, Giannis Daras, Sarah Pratt, Vivek Ramanujan, Yonatan Bitton, Kalyani Marathe, Stephen Mussmann, Richard Vencu, Mehdi Cherti, Ranjay Krishna, Pang Wei Koh, Olga Saukh, Alexander Ratner, Shuran Song, Hannaneh Hajishirzi, Ali Farhadi, Romain Beaumont, Sewoong Oh, Alex Dimakis, Jenia Jitsev, Yair Carmon, Vaishaal Shankar, Ludwig Schmidt

Optimal Learners for Realizable Regression: PAC Learning and Online Learning
Idan Attias, Steve Hanneke, Alkis Kalavasis, Amin Karbasi, Grigoris Velegkas

The Surprising Effectiveness of Diffusion Models for Optical Flow and Monocular Depth Estimation
Saurabh Saxena, Charles Herrmann, Junhwa Hur, Abhishek Kar, Mohammad Norouzi*, Deqing Sun, David J. Fleet


Journal Track

Graph Clustering with Graph Neural Networks
Anton Tsitsulin, John Palowitch, Bryan Perozzi, Emmanuel Müller


Spotlight Papers

Alternating Updates for Efficient Transformers (see blog post)
Cenk Baykal, Dylan Cutler, Nishanth Dikkala, Nikhil Ghosh*, Rina Panigrahy, Xin Wang

Does Localization Inform Editing? Surprising Differences in Causality-Based Localization vs. Knowledge Editing in Language Models
Peter Hase, Mohit Bansal, Been Kim, Asma Ghandeharioun

Is Learning in Games Good for the Learners?
William Brown, Jon Schneider, Kiran Vodrahalli

Participatory Personalization in Classification
Hailey Joren, Chirag Nagpal, Katherine Heller, Berk Ustun

Tight Risk Bounds for Gradient Descent on Separable Data
Matan Schliserman, Tomer Koren

Counterfactual Memorization in Neural Language Models
Chiyuan Zhang, Daphne Ippolito, Katherine Lee, Matthew Jagielski, Florian Tramèr, Nicholas Carlini

Debias Coarsely, Sample Conditionally: Statistical Downscaling through Optimal Transport and Probabilistic Diffusion Models
Zhong Yi Wan, Ricardo Baptista, Anudhyan Boral, Yi-Fan Chen, John Anderson, Fei Sha, Leonardo Zepeda-Nunez

Faster Margin Maximization Rates for Generic Optimization Methods
Guanghui Wang, Zihao Hu, Vidya Muthukumar, Jacob Abernethy

From Pixels to UI Actions: Learning to Follow Instructions via Graphical User Interfaces
Peter Shaw, Mandar Joshi, James Cohan, Jonathan Berant, Panupong Pasupat, Hexiang Hu, Urvashi Khandelwal, Kenton Lee, Kristina N Toutanova

PAC Learning Linear Thresholds from Label Proportions
Anand Brahmbhatt, Rishi Saket, Aravindan Raghuveer

SPAE: Semantic Pyramid AutoEncoder for Multimodal Generation with Frozen LLMs
Lijun Yu*, Yong Cheng, Zhiruo Wang, Vivek Kumar, Wolfgang Macherey, Yanping Huang, David Ross, Irfan Essa, Yonatan Bisk, Ming-Hsuan Yang, Kevin Murphy, Alexander Hauptmann, Lu Jiang

Adaptive Data Analysis in a Balanced Adversarial Model
Kobbi Nissim, Uri Stemmer, Eliad Tsfadia

Lexinvariant Language Models
Qian Huang, Eric Zelikman, Sarah Chen, Yuhuai Wu, Gregory Valiant, Percy Liang

On Quantum Backpropagation, Information Reuse, and Cheating Measurement Collapse
Amira Abbas, Robbie King, Hsin-Yuan Huang, William J. Huggins, Ramis Movassagh, Dar Gilboa, Jarrod McClean

Private Estimation Algorithms for Stochastic Block Models and Mixture Models
Hongjie Chen, Vincent Cohen-Addad, Tommaso d’Orsi, Alessandro Epasto, Jacob Imola, David Steurer, Stefan Tiegel

Provably Fast Finite Particle Variants of SVGD via Virtual Particle Stochastic Approximation
Aniket Das, Dheeraj Nagaraj

Private (Stochastic) Non-Convex Optimization Revisited: Second-Order Stationary Points and Excess Risks
Arun Ganesh, Daogao Liu*, Sewoong Oh, Abhradeep Guha Thakurta

Uncovering the Hidden Dynamics of Video Self-supervised Learning under Distribution Shifts
Pritam Sarkar, Ahmad Beirami, Ali Etemad

AIMS: All-Inclusive Multi-Level Segmentation for Anything
Lu Qi, Jason Kuen, Weidong Guo, Jiuxiang Gu, Zhe Lin, Bo Du, Yu Xu, Ming-Hsuan Yang

DreamHuman: Animatable 3D Avatars from Text
Nikos Kolotouros, Thiemo Alldieck, Andrei Zanfir, Eduard Gabriel Bazavan, Mihai Fieraru, Cristian Sminchisescu

Follow-ups Also Matter: Improving Contextual Bandits via Post-serving Contexts
Chaoqi Wang, Ziyu Ye, Zhe Feng, Ashwinkumar Badanidiyuru, Haifeng Xu

Learning List-Level Domain-Invariant Representations for Ranking
Ruicheng Xian*, Honglei Zhuang, Zhen Qin, Hamed Zamani*, Jing Lu, Ji Ma, Kai Hui, Han Zhao, Xuanhui Wang, Michael Bendersky

Optimal Guarantees for Algorithmic Reproducibility and Gradient Complexity in Convex Optimization
Liang Zhang, Junchi Yang, Amin Karbasi, Niao He

Unified Embedding: Battle-Tested Feature Representations for Web-Scale ML Systems
Benjamin Coleman, Wang-Cheng Kang, Matthew Fahrbach, Ruoxi Wang, Lichan Hong, Ed Chi, Derek Cheng

Proximity-Informed Calibration for Deep Neural Networks
Miao Xiong, Ailin Deng, Pang Wei Koh, Jiaying Wu, Shen Li, Jianqing Xu, Bryan Hooi


Papers

Anonymous Learning via Look-Alike Clustering: A Precise Analysis of Model Generalization
Adel Javanmard, Vahab Mirrokni

Better Private Linear Regression Through Better Private Feature Selection
Travis Dick, Jennifer Gillenwater*, Matthew Joseph

Binarized Neural Machine Translation
Yichi Zhang, Ankush Garg, Yuan Cao, Łukasz Lew, Behrooz Ghorbani*, Zhiru Zhang, Orhan Firat

BoardgameQA: A Dataset for Natural Language Reasoning with Contradictory Information
Mehran Kazemi, Quan Yuan, Deepti Bhatia, Najoung Kim, Xin Xu, Vaiva Imbrasaite, Deepak Ramachandran

Boosting with Tempered Exponential Measures
Richard Nock, Ehsan Amid, Manfred Warmuth

Concept Algebra for (Score-Based) Text-Controlled Generative Models
Zihao Wang, Lin Gui, Jeffrey Negrea, Victor Veitch

Deep Contract Design via Discontinuous Networks
Tonghan Wang, Paul Dütting, Dmitry Ivanov, Inbal Talgam-Cohen, David C. Parkes

Diffusion-SS3D: Diffusion Model for Semi-supervised 3D Object Detection
Cheng-Ju Ho, Chen-Hsuan Tai, Yen-Yu Lin, Ming-Hsuan Yang, Yi-Hsuan Tsai

Eliciting User Preferences for Personalized Multi-Objective Decision Making through Comparative Feedback
Han Shao, Lee Cohen, Avrim Blum, Yishay Mansour, Aadirupa Saha, Matthew Walter

Gradient Descent with Linearly Correlated Noise: Theory and Applications to Differential Privacy
Anastasia Koloskova*, Ryan McKenna, Zachary Charles, J Keith Rush, Hugh Brendan McMahan

Hardness of Low Rank Approximation of Entrywise Transformed Matrix Products
Tamas Sarlos, Xingyou Song, David P. Woodruff, Qiuyi (Richard) Zhang

Module-wise Adaptive Distillation for Multimodality Foundation Models

Chen Liang, Jiahui Yu, Ming-Hsuan Yang, Matthew Brown, Yin Cui, Tuo Zhao, Boqing Gong, Tianyi Zhou

Multi-Swap k-Means++
Lorenzo Beretta, Vincent Cohen-Addad, Silvio Lattanzi, Nikos Parotsidis

OpenMask3D: Open-Vocabulary 3D Instance Segmentation
Ayça Takmaz, Elisabetta Fedele, Robert Sumner, Marc Pollefeys, Federico Tombari, Francis Engelmann

Order Matters in the Presence of Dataset Imbalance for Multilingual Learning
Dami Choi*, Derrick Xin, Hamid Dadkhahi, Justin Gilmer, Ankush Garg, Orhan Firat, Chih-Kuan Yeh, Andrew M. Dai, Behrooz Ghorbani

PopSign ASL v1.0: An Isolated American Sign Language Dataset Collected via Smartphones
Thad Starner, Sean Forbes, Matthew So, David Martin, Rohit Sridhar, Gururaj Deshpande, Sam Sepah, Sahir Shahryar, Khushi Bhardwaj, Tyler Kwok, Daksh Sehgal, Saad Hassan, Bill Neubauer, Sofia Vempala, Alec Tan, Jocelyn Heath, Unnathi Kumar, Priyanka Mosur, Tavenner Hall, Rajandeep Singh, Christopher Cui, Glenn Cameron, Sohier Dane, Garrett Tanzer

Semi-Implicit Denoising Diffusion Models (SIDDMs)
Yanwu Xu*, Mingming Gong, Shaoan Xie, Wei Wei, Matthias Grundmann, Kayhan Batmanghelich, Tingbo Hou

State2Explanation: Concept-Based Explanations to Benefit Agent Learning and User Understanding
Devleena Das, Sonia Chernova, Been Kim

StoryBench: A Multifaceted Benchmark for Continuous Story Visualization
Emanuele Bugliarello*, Hernan Moraldo, Ruben Villegas, Mohammad Babaeizadeh, Mohammad Taghi Saffar, Han Zhang, Dumitru Erhan, Vittorio Ferrari, Pieter-Jan Kindermans, Paul Voigtlaender

Subject-driven Text-to-Image Generation via Apprenticeship Learning
Wenhu Chen, Hexiang Hu, Yandong Li, Nataniel Ruiz, Xuhui Jia, Ming-Wei Chang, William W. Cohen

TpuGraphs: A Performance Prediction Dataset on Large Tensor Computational Graphs
Phitchaya Mangpo Phothilimthana, Sami Abu-El-Haija, Kaidi Cao*, Bahare Fatemi, Mike Burrows, Charith Mendis*, Bryan Perozzi

Training Chain-of-Thought via Latent-Variable Inference
Du Phan, Matthew D. Hoffman, David Dohan*, Sholto Douglas, Tuan Anh Le, Aaron Parisi, Pavel Sountsov, Charles Sutton, Sharad Vikram, Rif A. Saurous

Unified Lower Bounds for Interactive High-dimensional Estimation under Information Constraints
Jayadev Acharya, Clement L. Canonne, Ziteng Sun, Himanshu Tyagi

What You See is What You Read? Improving Text-Image Alignment Evaluation
Michal Yarom, Yonatan Bitton, Soravit Changpinyo, Roee Aharoni, Jonathan Herzig, Oran Lang, Eran Ofek, Idan Szpektor

When Does Confidence-Based Cascade Deferral Suffice?
Wittawat Jitkrittum, Neha Gupta, Aditya Krishna Menon, Harikrishna Narasimhan, Ankit Singh Rawat, Sanjiv Kumar

Accelerating Molecular Graph Neural Networks via Knowledge Distillation
Filip Ekström Kelvinius, Dimitar Georgiev, Artur Petrov Toshev, Johannes Gasteiger

AVIS: Autonomous Visual Information Seeking with Large Language Model Agent
Ziniu Hu*, Ahmet Iscen, Chen Sun, Kai-Wei Chang, Yizhou Sun, David Ross, Cordelia Schmid, Alireza Fathi

Beyond Invariance: Test-Time Label-Shift Adaptation for Addressing "Spurious" Correlations
Qingyao Sun, Kevin Patrick Murphy, Sayna Ebrahimi, Alexander D'Amour

Collaborative Score Distillation for Consistent Visual Editing
Subin Kim, Kyungmin Lee, June Suk Choi, Jongheon Jeong, Kihyuk Sohn, Jinwoo Shin

CommonScenes: Generating Commonsense 3D Indoor Scenes with Scene Graphs
Guangyao Zhai, Evin Pınar Örnek, Shun-Cheng Wu, Yan Di, Federico Tombari, Nassir Navab, Benjamin Busam

Computational Complexity of Learning Neural Networks: Smoothness and Degeneracy
Amit Daniely, Nathan Srebro, Gal Vardi

A Computationally Efficient Sparsified Online Newton Method
Fnu Devvrit*, Sai Surya Duvvuri, Rohan Anil, Vineet Gupta, Cho-Jui Hsieh, Inderjit S Dhillon

DDF-HO: Hand-Held Object Reconstruction via Conditional Directed Distance Field
Chenyangguang Zhang, Yan Di, Ruida Zhang, Guangyao Zhai, Fabian Manhardt, Federico Tombari, Xiangyang Ji

Double Auctions with Two-sided Bandit Feedback
Soumya Basu, Abishek Sankararaman

Grammar Prompting for Domain-Specific Language Generation with Large Language Models
Bailin Wang, Zi Wang, Xuezhi Wang, Yuan Cao, Rif A. Saurous, Yoon Kim

Inconsistency, Instability, and Generalization Gap of Deep Neural Network Training
Rie Johnson, Tong Zhang*

Large Graph Property Prediction via Graph Segment Training
Kaidi Cao*, Phitchaya Mangpo Phothilimthana, Sami Abu-El-Haija, Dustin Zelle, Yanqi Zhou, Charith Mendis*, Jure Leskovec, Bryan Perozzi

On Computing Pairwise Statistics with Local Differential Privacy
Badih Ghazi, Pritish Kamath, Ravi Kumar, Pasin Manurangsi, Adam Sealfon

On Student-teacher Deviations in Distillation: Does it Pay to Disobey?
Vaishnavh Nagarajan, Aditya Krishna Menon, Srinadh Bhojanapalli, Hossein Mobahi, Sanjiv Kumar

Optimal Cross-learning for Contextual Bandits with Unknown Context Distributions
Jon Schneider, Julian Zimmert

Near-Optimal k-Clustering in the Sliding Window Model
David Woodruff, Peilin Zhong, Samson Zhou

Post Hoc Explanations of Language Models Can Improve Language Models
Satyapriya Krishna, Jiaqi Ma, Dylan Z Slack, Asma Ghandeharioun, Sameer Singh, Himabindu Lakkaraju

Recommender Systems with Generative Retrieval
Shashank Rajput*, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Q. Tran, Jonah Samost, Maciej Kula, Ed H. Chi, Maheswaran Sathiamoorthy

Reinforcement Learning for Fine-tuning Text-to-Image Diffusion Models
Ying Fan, Olivia Watkins, Yuqing Du, Hao Liu, Moonkyung Ryu, Craig Boutilier, Pieter Abbeel, Mohammad Ghavamzadeh*, Kangwook Lee, Kimin Lee*

Replicable Clustering
Hossein Esfandiari, Amin Karbasi, Vahab Mirrokni, Grigoris Velegkas, Felix Zhou

Replicability in Reinforcement Learning
Amin Karbasi, Grigoris Velegkas, Lin Yang, Felix Zhou

Riemannian Projection-free Online Learning
Zihao Hu, Guanghui Wang, Jacob Abernethy

Sharpness-Aware Minimization Leads to Low-Rank Features
Maksym Andriushchenko, Dara Bahri, Hossein Mobahi, Nicolas Flammarion

What is the Inductive Bias of Flatness Regularization? A Study of Deep Matrix Factorization Models
Khashayar Gatmiry, Zhiyuan Li, Ching-Yao Chuang, Sashank Reddi, Tengyu Ma, Stefanie Jegelka

Block Low-Rank Preconditioner with Shared Basis for Stochastic Optimization
Jui-Nan Yen, Sai Surya Duvvuri, Inderjit S Dhillon, Cho-Jui Hsieh

Blocked Collaborative Bandits: Online Collaborative Filtering with Per-Item Budget Constraints
Soumyabrata Pal, Arun Sai Suggala, Karthikeyan Shanmugam, Prateek Jain

Boundary Guided Learning-Free Semantic Control with Diffusion Models
Ye Zhu, Yu Wu, Zhiwei Deng, Olga Russakovsky, Yan Yan

Conditional Adapters: Parameter-efficient Transfer Learning with Fast Inference
Tao Lei, Junwen Bai, Siddhartha Brahma, Joshua Ainslie, Kenton Lee, Yanqi Zhou, Nan Du*, Vincent Y. Zhao, Yuexin Wu, Bo Li, Yu Zhang, Ming-Wei Chang

Conformal Prediction for Time Series with Modern Hopfield Networks
Andreas Auer, Martin Gauch, Daniel Klotz, Sepp Hochreiter

Does Visual Pretraining Help End-to-End Reasoning?
Chen Sun, Calvin Luo, Xingyi Zhou, Anurag Arnab, Cordelia Schmid

Effective Robustness Against Natural Distribution Shifts for Models with Different Training Data
Zhouxing Shi*, Nicholas Carlini, Ananth Balashankar, Ludwig Schmidt, Cho-Jui Hsieh, Alex Beutel*, Yao Qin

Improving Neural Network Representations Using Human Similarity Judgments
Lukas Muttenthaler*, Lorenz Linhardt, Jonas Dippel, Robert A. Vandermeulen, Katherine Hermann, Andrew K. Lampinen, Simon Kornblith

Label Robust and Differentially Private Linear Regression: Computational and Statistical Efficiency
Xiyang Liu, Prateek Jain, Weihao Kong, Sewoong Oh, Arun Sai Suggala

Mnemosyne: Learning to Train Transformers with Transformers
Deepali Jain, Krzysztof Choromanski, Avinava Dubey, Sumeet Singh, Vikas Sindhwani, Tingnan Zhang, Jie Tan

Nash Regret Guarantees for Linear Bandits
Ayush Sawarni, Soumyabrata Pal, Siddharth Barman

A Near-Linear Time Algorithm for the Chamfer Distance
Ainesh Bakshi, Piotr Indyk, Rajesh Jayaram, Sandeep Silwal, Erik Waingarten.

On Differentially Private Sampling from Gaussian and Product Distributions
Badih Ghazi, Xiao Hu*, Ravi Kumar, Pasin Manurangsi

On Dynamic Programming Decompositions of Static Risk Measures in Markov Decision Processes
Jia Lin Hau, Erick Delage, Mohammad Ghavamzadeh*, Marek Petrik

ResMem: Learn What You Can and Memorize the Rest
Zitong Yang, Michal Lukasik, Vaishnavh Nagarajan, Zonglin Li, Ankit Singh Rawat, Manzil Zaheer, Aditya Krishna Menon, Sanjiv Kumar

Responsible AI (RAI) Games and Ensembles
Yash Gupta, Runtian Zhai, Arun Suggala, Pradeep Ravikumar

RoboCLIP: One Demonstration Is Enough to Learn Robot Policies
Sumedh A Sontakke, Jesse Zhang, Sébastien M. R. Arnold, Karl Pertsch, Erdem Biyik, Dorsa Sadigh, Chelsea Finn, Laurent Itti

Robust Concept Erasure via Kernelized Rate-Distortion Maximization
Somnath Basu Roy Chowdhury, Nicholas Monath, Kumar Avinava Dubey, Amr Ahmed, Snigdha Chaturvedi

Robust Multi-Agent Reinforcement Learning via Adversarial Regularization: Theoretical Foundation and Stable Algorithms
Alexander Bukharin, Yan Li, Yue Yu, Qingru Zhang, Zhehui Chen, Simiao Zuo, Chao Zhang, Songan Zhang, Tuo Zhao

Simplicity Bias in 1-Hidden Layer Neural Networks
Depen Morwani*, Jatin Batra, Prateek Jain, Praneeth Netrapalli

SLaM: Student-Label Mixing for Distillation with Unlabeled Examples
Vasilis Kontonis, Fotis Iliopoulos, Khoa Trinh, Cenk Baykal, Gaurav Menghani, Erik Vee

SNAP: Self-Supervised Neural Maps for Visual Positioning and Semantic Understanding
Paul-Edouard Sarlin*, Eduard Trulls, Marc Pollefeys, Jan Hosang, Simon Lynen

SOAR: Improved Indexing for Approximate Nearest Neighbor Search
Philip Sun, David Simcha, Dave Dopson, Ruiqi Guo, Sanjiv Kumar

StyleDrop: Text-to-Image Synthesis of Any Style
Kihyuk Sohn, Lu Jiang, Jarred Barber, Kimin Lee*, Nataniel Ruiz, Dilip Krishnan, Huiwen Chang*, Yuanzhen Li, Irfan Essa, Michael Rubinstein, Yuan Hao, Glenn Entis, Irina Blok, Daniel Castro Chin

Three Towers: Flexible Contrastive Learning with Pretrained Image Models
Jannik Kossen*, Mark Collier, Basil Mustafa, Xiao Wang, Xiaohua Zhai, Lucas Beyer, Andreas Steiner, Jesse Berent, Rodolphe Jenatton, Efi Kokiopoulou

Two-Stage Learning to Defer with Multiple Experts
Anqi Mao, Christopher Mohri, Mehryar Mohri, Yutao Zhong

AdANNS: A Framework for Adaptive Semantic Search
Aniket Rege, Aditya Kusupati, Sharan Ranjit S, Alan Fan, Qingqing Cao, Sham Kakade, Prateek Jain, Ali Farhadi

Cappy: Outperforming and Boosting Large Multi-Task LMs with a Small Scorer
Bowen Tan*, Yun Zhu, Lijuan Liu, Eric Xing, Zhiting Hu, Jindong Chen

Causal-structure Driven Augmentations for Text OOD Generalization
Amir Feder, Yoav Wald, Claudia Shi, Suchi Saria, David Blei

Dense-Exponential Random Features: Sharp Positive Estimators of the Gaussian Kernel
Valerii Likhosherstov, Krzysztof Choromanski, Avinava Dubey, Frederick Liu, Tamas Sarlos, Adrian Weller

Diffusion Hyperfeatures: Searching Through Time and Space for Semantic Correspondence
Grace Luo, Lisa Dunlap, Dong Huk Park, Aleksander Holynski, Trevor Darrell

Diffusion Self-Guidance for Controllable Image Generation
Dave Epstein, Allan Jabri, Ben Poole, Alexei A Efros, Aleksander Holynski

Fully Dynamic k-Clustering in Õ(k) Update Time
Sayan Bhattacharya, Martin Nicolas Costa, Silvio Lattanzi, Nikos Parotsidis

Improving CLIP Training with Language Rewrites
Lijie Fan, Dilip Krishnan, Phillip Isola, Dina Katabi, Yonglong Tian

LayoutGPT: Compositional Visual Planning and Generation with Large Language Models
Weixi Feng, Wanrong Zhu, Tsu-Jui Fu, Varun Jampani, Arjun Reddy Akula, Xuehai He, Sugato Basu, Xin Eric Wang, William Yang Wang

Offline Reinforcement Learning for Mixture-of-Expert Dialogue Management
Dhawal Gupta*, Yinlam Chow, Azamat Tulepbergenov, Mohammad Ghavamzadeh*, Craig Boutilier

Optimal Unbiased Randomizers for Regression with Label Differential Privacy
Ashwinkumar Badanidiyuru, Badih Ghazi, Pritish Kamath, Ravi Kumar, Ethan Jacob Leeman, Pasin Manurangsi, Avinash V Varadarajan, Chiyuan Zhang

Paraphrasing Evades Detectors of AI-generated Text, but Retrieval Is an Effective Defense
Kalpesh Krishna, Yixiao Song, Marzena Karpinska, John Wieting, Mohit Iyyer

ReMaX: Relaxing for Better Training on Efficient Panoptic Segmentation
Shuyang Sun*, Weijun Wang, Qihang Yu*, Andrew Howard, Philip Torr, Liang-Chieh Chen*

Robust and Actively Secure Serverless Collaborative Learning
Nicholas Franzese, Adam Dziedzic, Christopher A. Choquette-Choo, Mark R. Thomas, Muhammad Ahmad Kaleem, Stephan Rabanser, Congyu Fang, Somesh Jha, Nicolas Papernot, Xiao Wang

SpecTr: Fast Speculative Decoding via Optimal Transport
Ziteng Sun, Ananda Theertha Suresh, Jae Hun Ro, Ahmad Beirami, Himanshu Jain, Felix Yu

Structured Prediction with Stronger Consistency Guarantees
Anqi Mao, Mehryar Mohri, Yutao Zhong

Affinity-Aware Graph Networks
Ameya Velingker, Ali Kemal Sinop, Ira Ktena, Petar Veličković, Sreenivas Gollapudi

ARTIC3D: Learning Robust Articulated 3D Shapes from Noisy Web Image Collections
Chun-Han Yao*, Amit Raj, Wei-Chih Hung, Yuanzhen Li, Michael Rubinstein, Ming-Hsuan Yang, Varun Jampani

Black-Box Differential Privacy for Interactive ML
Haim Kaplan, Yishay Mansour, Shay Moran, Kobbi Nissim, Uri Stemmer

Bypassing the Simulator: Near-Optimal Adversarial Linear Contextual Bandits
Haolin Liu, Chen-Yu Wei, Julian Zimmert

DaTaSeg: Taming a Universal Multi-Dataset Multi-Task Segmentation Model

Xiuye Gu, Yin Cui*, Jonathan Huang, Abdullah Rashwan, Xuan Yang, Xingyi Zhou, Golnaz Ghiasi, Weicheng Kuo, Huizhong Chen, Liang-Chieh Chen*, David Ross

Easy Learning from Label Proportions
Robert Busa-Fekete, Heejin Choi*, Travis Dick, Claudio Gentile, Andres Munoz Medina

Efficient Data Subset Selection to Generalize Training Across Models: Transductive and Inductive Networks
Eeshaan Jain, Tushar Nandy, Gaurav Aggarwal, Ashish Tendulkar, Rishabh Iyer, Abir De

Faster Differentially Private Convex Optimization via Second-Order Methods
Arun Ganesh, Mahdi Haghifam*, Thomas Steinke, Abhradeep Guha Thakurta

Finding Safe Zones of Markov Decision Processes Policies
Lee Cohen, Yishay Mansour, Michal Moshkovitz

Focused Transformer: Contrastive Training for Context Scaling
Szymon Tworkowski, Konrad Staniszewski, Mikołaj Pacek, Yuhuai Wu*, Henryk Michalewski, Piotr Miłoś

Front-door Adjustment Beyond Markov Equivalence with Limited Graph Knowledge
Abhin Shah, Karthikeyan Shanmugam, Murat Kocaoglu

H-Consistency Bounds: Characterization and Extensions
Anqi Mao, Mehryar Mohri, Yutao Zhong

Inverse Dynamics Pretraining Learns Good Representations for Multitask Imitation
David Brandfonbrener, Ofir Nachum, Joan Bruna

Most Neural Networks Are Almost Learnable
Amit Daniely, Nathan Srebro, Gal Vardi

Multiclass Boosting: Simple and Intuitive Weak Learning Criteria
Nataly Brukhim, Amit Daniely, Yishay Mansour, Shay Moran

NeRF Revisited: Fixing Quadrature Instability in Volume Rendering
Mikaela Angelina Uy, Kiyohiro Nakayama, Guandao Yang, Rahul Krishna Thomas, Leonidas Guibas, Ke Li

Privacy Amplification via Compression: Achieving the Optimal Privacy-Accuracy-Communication Trade-off in Distributed Mean Estimation
Wei-Ning Chen, Dan Song, Ayfer Ozgur, Peter Kairouz

Private Federated Frequency Estimation: Adapting to the Hardness of the Instance
Jingfeng Wu*, Wennan Zhu, Peter Kairouz, Vladimir Braverman

RETVec: Resilient and Efficient Text Vectorizer
Elie Bursztein, Marina Zhang, Owen Skipper Vallis, Xinyu Jia, Alexey Kurakin

Symbolic Discovery of Optimization Algorithms
Xiangning Chen*, Chen Liang, Da Huang, Esteban Real, Kaiyuan Wang, Hieu Pham, Xuanyi Dong, Thang Luong, Cho-Jui Hsieh, Yifeng Lu, Quoc V. Le

A Tale of Two Features: Stable Diffusion Complements DINO for Zero-Shot Semantic Correspondence
Junyi Zhang, Charles Herrmann, Junhwa Hur, Luisa F. Polania, Varun Jampani, Deqing Sun, Ming-Hsuan Yang

A Trichotomy for Transductive Online Learning
Steve Hanneke, Shay Moran, Jonathan Shafer

A Unified Fast Gradient Clipping Framework for DP-SGD
William Kong, Andres Munoz Medina

Unleashing the Power of Randomization in Auditing Differentially Private ML
Krishna Pillutla, Galen Andrew, Peter Kairouz, H. Brendan McMahan, Alina Oprea, Sewoong Oh

(Amplified) Banded Matrix Factorization: A unified approach to private training
Christopher A Choquette-Choo, Arun Ganesh, Ryan McKenna, H Brendan McMahan, Keith Rush, Abhradeep Guha Thakurta, Zheng Xu

Adversarial Resilience in Sequential Prediction via Abstention
Surbhi Goel, Steve Hanneke, Shay Moran, Abhishek Shetty

Alternating Gradient Descent and Mixture-of-Experts for Integrated Multimodal Perception
Hassan Akbari, Dan Kondratyuk, Yin Cui, Rachel Hornung, Huisheng Wang, Hartwig Adam

Android in the Wild: A Large-Scale Dataset for Android Device Control
Christopher Rawles, Alice Li, Daniel Rodriguez, Oriana Riva, Timothy Lillicrap

Benchmarking Robustness to Adversarial Image Obfuscations
Florian Stimberg, Ayan Chakrabarti, Chun-Ta Lu, Hussein Hazimeh, Otilia Stretcu, Wei Qiao, Yintao Liu, Merve Kaya, Cyrus Rashtchian, Ariel Fuxman, Mehmet Tek, Sven Gowal

Building Socio-culturally Inclusive Stereotype Resources with Community Engagement
Sunipa Dev, Jaya Goyal, Dinesh Tewari, Shachi Dave, Vinodkumar Prabhakaran

Consensus and Subjectivity of Skin Tone Annotation for ML Fairness
Candice Schumann, Gbolahan O Olanubi, Auriel Wright, Ellis Monk Jr*, Courtney Heldreth, Susanna Ricco

Counting Distinct Elements Under Person-Level Differential Privacy
Alexander Knop, Thomas Steinke

DICES Dataset: Diversity in Conversational AI Evaluation for Safety
Lora Aroyo, Alex S. Taylor, Mark Diaz, Christopher M. Homan, Alicia Parrish, Greg Serapio-García, Vinodkumar Prabhakaran, Ding Wang

Does Progress on ImageNet Transfer to Real-world Datasets?
Alex Fang, Simon Kornblith, Ludwig Schmidt

Estimating Generic 3D Room Structures from 2D Annotations
Denys Rozumnyi*, Stefan Popov, Kevis-kokitsi Maninis, Matthias Nießner, Vittorio Ferrari

Large Language Model as Attributed Training Data Generator: A Tale of Diversity and Bias
Yue Yu, Yuchen Zhuang, Jieyu Zhang, Yu Meng, Alexander Ratner, Ranjay Krishna, Jiaming Shen, Chao Zhang

MADLAD-400: A Multilingual And Document-Level Large Audited Dataset
Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat

Mechanic: A Learning Rate Tuner
Ashok Cutkosky, Aaron Defazio, Harsh Mehta

NAVI: Category-Agnostic Image Collections with High-Quality 3D Shape and Pose Annotations
Varun Jampani, Kevis-kokitsi Maninis, Andreas Engelhardt, Arjun Karpur, Karen Truong, Kyle Sargent, Stefan Popov, Andre Araujo, Ricardo Martin Brualla, Kaushal Patel, Daniel Vlasic, Vittorio Ferrari, Ameesh Makadia, Ce Liu*, Yuanzhen Li, Howard Zhou

Neural Ideal Large Eddy Simulation: Modeling Turbulence with Neural Stochastic Differential Equations
Anudhyan Boral, Zhong Yi Wan, Leonardo Zepeda-Nunez, James Lottes, Qing Wang, Yi-Fan Chen, John Roberts Anderson, Fei Sha

Restart Sampling for Improving Generative Processes
Yilun Xu, Mingyang Deng, Xiang Cheng, Yonglong Tian, Ziming Liu, Tommi Jaakkola

Rethinking Incentives in Recommender Systems: Are Monotone Rewards Always Beneficial?
Fan Yao, Chuanhao Li, Karthik Abinav Sankararaman, Yiming Liao, Yan Zhu, Qifan Wang, Hongning Wang, Haifeng Xu

Revisiting Evaluation Metrics for Semantic Segmentation: Optimization and Evaluation of Fine-grained Intersection over Union
Zifu Wang, Maxim Berman, Amal Rannen-Triki, Philip Torr, Devis Tuia, Tinne Tuytelaars, Luc Van Gool, Jiaqian Yu, Matthew B. Blaschko

RoboHive: A Unified Framework for Robot Learning
Vikash Kumar, Rutav Shah, Gaoyue Zhou, Vincent Moens, Vittorio Caggiano, Abhishek Gupta, Aravind Rajeswaran

SatBird: Bird Species Distribution Modeling with Remote Sensing and Citizen Science Data
Mélisande Teng, Amna Elmustafa, Benjamin Akera, Yoshua Bengio, Hager Radi, Hugo Larochelle, David Rolnick

Sparsity-Preserving Differentially Private Training of Large Embedding Models
Badih Ghazi, Yangsibo Huang*, Pritish Kamath, Ravi Kumar, Pasin Manurangsi, Amer Sinha, Chiyuan Zhang

StableRep: Synthetic Images from Text-to-Image Models Make Strong Visual Representation Learners
Yonglong Tian, Lijie Fan, Phillip Isola, Huiwen Chang, Dilip Krishnan

Towards Federated Foundation Models: Scalable Dataset Pipelines for Group-Structured Learning
Zachary Charles, Nicole Mitchell, Krishna Pillutla, Michael Reneer, Zachary Garrett

Universality and Limitations of Prompt Tuning
Yihan Wang, Jatin Chauhan, Wei Wang, Cho-Jui Hsieh

Unsupervised Semantic Correspondence Using Stable Diffusion
Eric Hedlin, Gopal Sharma, Shweta Mahajan, Hossam Isack, Abhishek Kar, Andrea Tagliasacchi, Kwang Moo Yi

YouTube-ASL: A Large-Scale, Open-Domain American Sign Language-English Parallel Corpus
Dave Uthus, Garrett Tanzer, Manfred Georg

The Noise Level in Linear Regression with Dependent Data
Ingvar Ziemann, Stephen Tu, George J. Pappas, Nikolai Matni

* Work done while at Google

Source: Google AI Blog


Sparsity-preserving differentially private training

Posted by Yangsibo Huang, Research Intern, Google Research; Chiyuan Zhang, Research Scientist, Google Research

Large embedding models have emerged as a fundamental tool for various applications in recommendation systems [1, 2] and natural language processing [3, 4, 5]. Such models enable the integration of non-numerical data into deep learning models by mapping categorical or string-valued input attributes with large vocabularies to fixed-length representation vectors using embedding layers. These models are widely deployed in personalized recommendation systems and achieve state-of-the-art performance in language tasks, such as language modeling, sentiment analysis, and question answering. In many such scenarios, privacy is an equally important feature when deploying those models. As a result, various techniques have been proposed to enable private data analysis. Among those, differential privacy (DP) is a widely adopted definition that limits exposure of individual user information while still allowing for the analysis of population-level patterns.

For training deep neural networks with DP guarantees, the most widely used algorithm is DP-SGD (DP stochastic gradient descent). One key component of DP-SGD is adding Gaussian noise to every coordinate of the gradient vectors during training. However, this creates scalability challenges when applied to large embedding models, because they rely on gradient sparsity for efficient training, but adding noise to all the coordinates destroys sparsity.

To mitigate this gradient sparsity problem, in “Sparsity-Preserving Differentially Private Training of Large Embedding Models” (to be presented at NeurIPS 2023), we propose a new algorithm called adaptive filtering-enabled sparse training (DP-AdaFEST). At a high level, the algorithm maintains the sparsity of the gradient by selecting only a subset of feature rows to which noise is added at each iteration. The key is to make such selections differentially private so that a three-way balance is achieved among the privacy cost, the training efficiency, and the model utility. Our empirical evaluation shows that DP-AdaFEST achieves a substantially sparser gradient, with a reduction in gradient size of over 105X compared to the dense gradient produced by standard DP-SGD, while maintaining comparable levels of accuracy. This gradient size reduction could translate into 20X wall-clock time improvement.


Overview

To better understand the challenges and our solutions to the gradient sparsity problem, let us start with an overview of how DP-SGD works during training. As illustrated by the figure below, DP-SGD operates by clipping the gradient contribution from each example in the current random subset of samples (called a mini-batch), and adding coordinate-wise Gaussian noise to the average gradient during each iteration of stochastic gradient descent (SGD). DP-SGD has demonstrated its effectiveness in protecting user privacy while maintaining model utility in a variety of applications [6, 7].

An illustration of how DP-SGD works. During each training step, a mini-batch of examples is sampled, and used to compute the per-example gradients. Those gradients are processed through clipping, aggregation and summation of Gaussian noise to produce the final privatized gradients.

The challenges of applying DP-SGD to large embedding models mainly come from 1) the non-numerical feature fields like user/product IDs and categories, and 2) words and tokens that are transformed into dense vectors through an embedding layer. Due to the vocabulary sizes of those features, the process requires large embedding tables with a substantial number of parameters. In contrast to the number of parameters, the gradient updates are usually extremely sparse because each mini-batch of examples only activates a tiny fraction of embedding rows (the figure below visualizes the ratio of zero-valued coordinates, i.e., the sparsity, of the gradients under various batch sizes). This sparsity is heavily leveraged for industrial applications that efficiently handle the training of large-scale embeddings. For example, Google Cloud TPUs, custom-designed AI accelerators that are optimized for training and inference of large AI models, have dedicated APIs to handle large embeddings with sparse updates. This leads to significantly improved training throughput compared to training on GPUs, which at this time did not have specialized optimization for sparse embedding lookups. On the other hand, DP-SGD completely destroys the gradient sparsity because it requires adding independent Gaussian noise to all the coordinates. This creates a road block for private training of large embedding models as the training efficiency would be significantly reduced compared to non-private training.

Embedding gradient sparsity (the fraction of zero-value gradient coordinates) in the Criteo pCTR model (see below). The figure reports the gradient sparsity, averaged over 50 update steps, of the top five categorical features (out of a total of 26) with the highest number of buckets, as well as the sparsity of all categorical features. The sprasity decreases with the batch size as more examples hit more rows in the embedding table, creating non-zero gradients. However, the sparsity is above 0.97 even for very large batch sizes. This pattern is consistently observed for all the five features.

Algorithm

Our algorithm is built by extending standard DP-SGD with an extra mechanism at each iteration to privately select the “hot features”, which are the features that are activated by multiple training examples in the current mini-batch. As illustrated below, the mechanism works in a few steps:

  1. Compute how many examples contributed to each feature bucket (we call each of the possible values of a categorical feature a “bucket”).
  2. Restrict the total contribution from each example by clipping their counts.
  3. Add Gaussian noise to the contribution count of each feature bucket.
  4. Select only the features to be included in the gradient update that have a count above a given threshold (a sparsity-controlling parameter), thus maintaining sparsity. This mechanism is differentially private, and the privacy cost can be easily computed by composing it with the standard DP-SGD iterations.
Illustration of the process of the algorithm on a synthetic categorical feature that has 20 buckets. We compute the number of examples contributing to each bucket, adjust the value based on per-example total contributions (including those to other features), add Gaussian noise, and retain only those buckets with a noisy contribution exceeding the threshold for (noisy) gradient update.

Theoretical motivation

We provide the theoretical motivation that underlies DP-AdaFEST by viewing it as optimization using stochastic gradient oracles. Standard analysis of stochastic gradient descent in a theoretical setting decomposes the test error of the model into “bias” and “variance” terms. The advantage of DP-AdaFEST can be viewed as reducing variance at the cost of slightly increasing the bias. This is because DP-AdaFEST adds noise to a smaller set of coordinates compared to DP-SGD, which adds noise to all the coordinates. On the other hand, DP-AdaFEST introduces some bias to the gradients since the gradient on the embedding features are dropped with some probability. We refer the interested reader to Section 3.4 of the paper for more details.


Experiments

We evaluate the effectiveness of our algorithm with large embedding model applications, on public datasets, including one ad prediction dataset (Criteo-Kaggle) and one language understanding dataset (SST-2). We use DP-SGD with exponential selection as a baseline comparison.

The effectiveness of DP-AdaFEST is evident in the figure below, where it achieves significantly higher gradient size reduction (i.e., gradient sparsity) than the baseline while maintaining the same level of utility (i.e., only minimal performance degradation).

Specifically, on the Criteo-Kaggle dataset, DP-AdaFEST reduces the gradient computation cost of regular DP-SGD by more than 5x105 times while maintaining a comparable AUC (which we define as a loss of less than 0.005). This reduction translates into a more efficient and cost-effective training process. In comparison, as shown by the green line below, the baseline method is not able to achieve reasonable cost reduction within such a small utility loss threshold.

In language tasks, there isn't as much potential for reducing the size of gradients, because the vocabulary used is often smaller and already quite compact (shown on the right below). However, the adoption of sparsity-preserving DP-SGD effectively obviates the dense gradient computation. Furthermore, in line with the bias-variance trade-off presented in the theoretical analysis, we note that DP-AdaFEST occasionally exhibits superior utility compared to DP-SGD when the reduction in gradient size is minimal. Conversely, when incorporating sparsity, the baseline algorithm faces challenges in maintaining utility.

A comparison of the best gradient size reduction (the ratio of the non-zero gradient value counts between regular DP-SGD and sparsity-preserving algorithms) achieved under ε =1.0 by DP-AdaFEST (our algorithm) and the baseline algorithm (DP-SGD with exponential selection) compared to DP-SGD at different thresholds for utility difference. A higher curve indicates a better utility/efficiency trade-off.

In practice, most ad prediction models are being continuously trained and evaluated. To simulate this online learning setup, we also evaluate with time-series data, which are notoriously challenging due to being non-stationary. Our evaluation uses the Criteo-1TB dataset, which comprises real-world user-click data collected over 24 days. Consistently, DP-AdaFEST reduces the gradient computation cost of regular DP-SGD by more than 104 times while maintaining a comparable AUC.

A comparison of the best gradient size reduction achieved under ε =1.0 by DP-AdaFEST (our algorithm) and DP-SGD with exponential selection (a previous algorithm) compared to DP-SGD at different thresholds for utility difference. A higher curve indicates a better utility/efficiency trade-off. DP-AdaFEST consistently outperforms the previous method.

Conclusion

We present a new algorithm, DP-AdaFEST, for preserving gradient sparsity in differentially private training — particularly in applications involving large embedding models, a fundamental tool for various applications in recommendation systems and natural language processing. Our algorithm achieves significant reductions in gradient size while maintaining accuracy on real-world benchmark datasets. Moreover, it offers flexible options for balancing utility and efficiency via sparsity-controlling parameters, while our proposals offer much better privacy-utility loss.


Acknowledgements

This work was a collaboration with Badih Ghazi, Pritish Kamath, Ravi Kumar, Pasin Manurangsi and Amer Sinha.

Source: Google AI Blog