Google TV released the top movies and shows from 2023 in collections like “most-watched movies” “most-watchlisted,” and others.
.png)
Systems such as Gmail, YouTube and Google Play rely on text classification models to identify harmful content including phishing attacks, inappropriate comments, and scams. These types of texts are harder for machine learning models to classify because bad actors rely on adversarial text manipulations to actively attempt to evade the classifiers. For example, they will use homoglyphs, invisible characters, and keyword stuffing to bypass defenses.
To help make text classifiers more robust and efficient, we’ve developed a novel, multilingual text vectorizer called RETVec (Resilient & Efficient Text Vectorizer) that helps models achieve state-of-the-art classification performance and drastically reduces computational cost. Today, we’re sharing how RETVec has been used to help protect Gmail inboxes.
Strengthening the Gmail Spam Classifier with RETVec
Figure 1. RETVec-based Gmail Spam filter improvements.
Over the past year, we battle-tested RETVec extensively inside Google to evaluate its usefulness and found it to be highly effective for security and anti-abuse applications. In particular, replacing the Gmail spam classifier’s previous text vectorizer with RETVec allowed us to improve the spam detection rate over the baseline by 38% and reduce the false positive rate by 19.4%. Additionally, using RETVec reduced the TPU usage of the model by 83%, making the RETVec deployment one of the largest defense upgrades in recent years. RETVec achieves these improvements by sporting a very lightweight word embedding model (~200k parameters), allowing us to reduce the Transformer model’s size at equal or better performance, and having the ability to split the computation between the host and TPU in a network and memory efficient manner.
RETVec Benefits
RETVec achieves these improvements by combining a novel, highly-compact character encoder, an augmentation-driven training regime, and the use of metric learning. The architecture details and benchmark evaluations are available in our NeurIPS 2023 paper and we open-source RETVec on Github.
Due to its novel architecture, RETVec works out-of-the-box on every language and all UTF-8 characters without the need for text preprocessing, making it the ideal candidate for on-device, web, and large-scale text classification deployments. Models trained with RETVec exhibit faster inference speed due to its compact representation. Having smaller models reduces computational costs and decreases latency, which is critical for large-scale applications and on-device models.
Figure 1. RETVec architecture diagram.
Models trained with RETVec can be seamlessly converted to TFLite for mobile and edge devices, as a result of a native implementation in TensorFlow Text. For web application model deployment, we provide a TensorflowJS layer implementation that is available on Github and you can check out a demo web page running a RETVec-based model.
Figure 2. Typo resilience of text classification models trained from scratch using different vectorizers.
RETVec is a novel open-source text vectorizer that allows you to build more resilient and efficient server-side and on-device text classifiers. The Gmail spam filter uses it to help protect Gmail inboxes against malicious emails.
If you would like to use RETVec for your own use cases or research, we created a tutorial to help you get started.
This research was conducted by Elie Bursztein, Marina Zhang, Owen Vallis, Xinyu Jia, and Alexey Kurakin. We would like to thank Gengxin Miao, Brunno Attorre, Venkat Sreepati, Lidor Avigad, Dan Givol, Rishabh Seth and Melvin Montenegro and all the Googlers who contributed to the project.
Mira Lane explores how AI is empowering creativity with pioneering artist Refik Anadol.
| Version | Planned Release Type* |
Projected launch* | Projected sunset* |
|---|---|---|---|
| V16 | Major | January/February 2024 | January 2025 |
| V16.1 | Minor | April/May 2024 | January 2025 |
| V17 | Major | May/June 2024 | May 2025 |
| V17.1 | Minor | July/August 2024 | May 2025 |
| V18 | Major | September/October 2024 | September 2025 |
- David Stevens, Google Ads API Product Manager
Two years ago, Junior Achievement of Greater Kansas City chose Google Fiber’s 2 Gig service and we’ve never looked back. In recognition of National Entrepreneurship Month, we’re highlighting how Junior Achievement of Greater Kansas City uses reliable, high-speed internet to help achieve our goal of empowering the next generation of professionals, leaders, and entrepreneurs.
.jpg)
.jpg "Torey Berndt, Director of Marketing and Communications, Junior Achievement of Greater Kansas City")
We’re announcing Google Play’s Best of 2023 awards, celebrating the best apps, games and books of the year.
Hello everyone,
Since joining the team at the start of this year, I’ve been continuously inspired by our incredible community of people building apps and games, and it’s been my privilege to support your hard work and creativity. In particular, it’s been exciting to see you not only identify new user needs, but to develop innovative ways of solving them with your apps and games. This year, our annual Best of Play awards introduced new categories to recognize these achievements, including “Best with AI” and “Best multi-device app and game.”
No matter where you are on your developer journey, whether you just published your first app or already reached a global audience, Google Play is committed to being your partner in growth. Here’s a look back at some of the key tools, features, and programs we built this year to help you reach your full potential and build a successful business on Google Play.
Check out the video below or keep reading for more details about this year’s updates.
This year, our Indie Games programs helped businesses of all sizes grow on Play. Through our Indie Games Fund, we awarded $2 million and offered hands-on support to 10 Latin American studios to help them grow their games on our platform. And for mentorship from Google and industry experts, early-career indie developers can apply for the Indie Games Accelerator from now until December 12, 2023.
We’re also pushing for greater representation and equity in the developer community by giving $600,000 across 13 nonprofit organizations around the world to support more inclusive programming. For example, we’ve partnered with Global Game Jam and the IGDA Foundation to host a game jam that helps women across Asia and Latin America launch careers in game development.
And finally, we continued our tour of the world by sharing and celebrating your stories through #WeArePlay. This year, we spotlighted over 260 app and game businesses from Europe, Japan, India and more.
I’m excited about the future of Google Play, which will see even more updates and improvements to the work we did this year.
For example, you may have noticed that we gave the Play Store a fresh look this year, with more visual components, new video capabilities, and fluid animations, and introduced new ways for people to discover, engage, and re-engage with your apps and games. In addition to new original editorial content and livestream events, we also developed new ways for users to be rewarded for their play, such as with time-bound offers and promotions.
Next year, we’ll build on that investment, going deeper to provide more value to users — and, in turn, to your businesses. We’re looking to:
As you know, Google Play is more than an app marketplace. We connect people with the experiences they’ll love, wherever they are, on whatever device they’re using. A big part of this is our continued focus on making it easier for people to find your latest, most relevant app and game content. It also means going beyond the Play Store to deliver this content to people across their devices, when and where it's most relevant.
Once again, I want to thank you for all the hard work you’ve put into making Google Play the best place for apps and games. I can’t wait to see what you build next.
On behalf of all of us at Google Play, happy holidays and best wishes for an amazing 2024.
Google launches a new cybersecurity hub in Málaga, Spain.
Google launches new $10 million program to advance cybersecurity skills development and training and helping local community organizations.
Posted by Sam Bright, Vice President & General Manager, Google Play