Google, large-scale security update incorporating artificial intelligence technology into Gmail
Haebom
Gmail has released a new AI spam detection feature called "RETVec." It has been tested internally for over a year and is now being applied to Gmail accounts. The technology is an abbreviation for Resilient & Efficient Text Vectorizer and is said to have been created to deal with adversarial text. RETVec boasts the following features:
Understands and filters emails containing special characters, emojis, typos, and other miscellaneous characters. (Can efficiently encode all UTF-8 characters and words.)
Supports over 100 languages, no need for separate lookup tables or fixed vocabulary sizes.
Reduces resource usage compared to alternative approaches that use fixed vocabulary sizes or synonym tables.
Actually, if you look at it this way, you might wonder, what's so good about this? This is the change that users will actually feel.
Improved spam detection rate: By replacing the existing text vectorizer with RETVec, the spam detection rate is improved by 38% compared to the baseline.
Reduced false positive rate: 19.4% reduction in spam creation.
Reduced resource usage: Reduced TPU usage of models by 83%.
According to Google's announcement, RETVec has been released as open source and can be applied to various areas such as Google Maps local comments and store review sections in the future. For more information, please refer to the Google Security Blog.
📚 Welcome to Haebom's archives. --- I post articles related to IT 💻, economy 💰, and humanities 🎭. If you are curious about my thoughts, perspectives or interests, please subscribe. Would you like to be notified when new articles are posted? 🔔 Yes, that means subscribe. haebom@kakao.com
