Last month, Google has connected its TensorFlow machine learning system to develop additional spam filters for Gmail users. Now the service blocks more messages every day.
If we consider that now Gmail has more than 1 billion users, then this is not such a huge effect. However, new filters are addition to existing methods. According to representatives of the company, thanks to the innovation, Gmail now blocks 99.9% spam, since the remaining percentage share is too difficult to determine.
The service has long been using artificial intelligence in its work, but the previous algorithms were determined only by obvious spam, and machine learning forms new templates to determine which email does not deserve confidence. Algorithms created in this way compare a huge number of metrics, ranging from formatting a letter and ending with the time to send it.
Google claims that the integration of TensorFlow in the post office will also allow you to better personalize spam filters. This process occurs over the years, since Gmail analyzes certain signals from users about what they consider spam, and turns them into ready-made models.
Artificial intelligence begins to penetrate all spheres of our life, and large data and machine learning have