Gmail uses TensorFlow machine learning to remove 100 million spam messages every day

Google has announced using its new protections powered TensorFlow open source machine learning framework is blocking around 100 million additional spam messages on Gmail every day. It is now blocking spam categories that used to be very hard to detect. TensorFlow was able to block image-based messages, emails with hidden embedded content, and messages from newly created domains that try to hide a low volume of spammy messages within legitimate traffic.

Machine learning makes catching spam possible as it helps in identifying patterns in large data sets that humans who create the rules might not catch. With millions of emails every day, applying ML at scale can be complex and time-consuming. TensorFlow includes many tools that make the ML process easier and more efficient.

TensorFlow also gives the flexibility to easily train and experiment with different models in parallel to develop the most effective approach, instead of running one experiment at a time. Thanks to its open nature, TensorFlow is used by teams and researchers all over the world. Google also clarified that it is currently experimenting with TensorFlow in other security-related areas, such as phishing and malware detection.