How applying machine learning in Gmail allowed to level the spam flow in the mail
Email is one of the main channels of business communication. Unfortunately, it also becomes a place of accumulation of spam – various mailings, letters from scammers and other unnecessary garbage. It is sometimes difficult to find or miss an important letter in this thread.
The world’s leading email service, Gmail, has recently stepped up efforts to combat this problem. It is reported that the system of the company with the use of machine learning now blocks almost 100% of spam.
Google’s machine learning application is based on advanced e-mail debris filtering algorithms.
Machine Learning Technology (MLT) TensorFlow is a proprietary Google development whose product is Gmail. The tool generates new, more powerful and advanced filters, which determines the spam character of the letter with high accuracy
Its integration into the mail client is already the second update of Gmail lately. At the beginning of the year, the service changed the appearance and expanded the functionality of its mobile application. Last but not least, thanks to MLT.
Representatives of Gmail report that now, after applying the innovation, Gmail is able to block up to one hundred million harmful messages a day. In percentage terms, the figure is even more impressive – it is about 99.9 e-mail garbage.
The mail client uses a neural network in conjunction with already working filters, which for many years have been eradicating spam. But their algorithms are no longer perfect, since they block only obvious spam. At the same time, artificial intelligence finds new patterns that indicate the potentially harmful nature of the letter. Despite the fact that the QS algorithms are based on a huge base of calculations and metrics, the level of elaboration of TensorFlow simplifies the management of this data.
Interestingly, Google launched its SMO back in 2015. It was originally a platform where developers could create tools based on artificial intelligence that solved a wide range of tasks.
The integration of TensorFlow in Gmail allows for better customization of spam filters. And, despite the fact that a single concept of spam does not exist, Google’s AI can find the most appropriate concept of what can be considered.
Gmail is an email service that provides access to electronic online mailboxes. All over the world, more than 400 million people use it. As of 2019, this is the most popular email client.