AI for anti spam !

MailCleaner is proud to announce the new Machine Learning module. In the past few years, groundbreaking advancements were made in Machine Learning thanks to Deep Neural Networks. The field of Natural Language Processing was particularly affected as these algorithms are able to build incredibly detailed representations of the human languages.

MailCleaner has integrated a Machine Learning module. At its core, the module is powered by multiple Deep Neural Networks implemented and optimized in house by the new MailCleaner data sciences team in close collaboration with our field experts. Specifically applied to a subset of the entire traffic, this module exhibits outstanding classification capabilities.

Fig.1 – ham and spam emails separated by the MailCleaner Machine Learning module. Hams are close to zero and spams close to 100

Fig.2 – False positive and false negative optimization depending on thresholds in the MailCleaner Machine Learning module.

Thanks to the abstraction power of Neural Networks, the MailCleaner Machine Learning module is able to adapt very quickly to the constantly evolving world of spam e-mails. By their nature, these types of filters are very difficult to reverse-engineer and their judgment is practically incorruptible. These properties make them very powerful and robust shields against the creativity of spammers who continually try to bypass or break anti spam filters.

The first version of the Machine Learning module efficiently consolidate and boost the already excellent filtering performance of MailCleaner. The boundaries of Machine Learning are being endlessly redefined, we are very excited to take advantage of its power and are already looking forward to announcing the next versions of this module.