Gmail improves spam filter with neural network, sender tools

Gmail product manager Sri Harsha Somanchi has outlined some new steps the free email service is taking to try to improve its spam detection on the official Gmail blog. First, Google is deploying a new service, Gmail Postmaster Tools, which allows "qualified high-volume senders" to analyze what they send and how it's treated. Senders will be able to see data on sending errors, spam reports, and their reputations. (If this sounds familiar, Microsoft offers a similar service for bulk senders in the form of Smart Network Data Services, or SNDS.)

Interestingly, Google claims to have implemented ideas from its Google Search and Google Now services to improve its machine learning processes for spam detection. The spam filter now employs an artificial neural network to help identify what the company calls "especially sneaky" spam that might otherwise appear to be legitimate mail.  The neural network should also help detect email impersonation tricks like phishing attempts. Somanchi claims the neural network can do all of this while remembering that some people actually enjoy stuff like promotional newsletters. Anything that cuts the number of times I'm disappointed to find out I didn't actually win the British lotto is a welcome change.

Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.