With Chrome machine learning, it captures phishing, enhances history, or hides notifications

Posted by


Google shows how machine learning integrates into practice in Chrome. Everything works locally. New model from March identifies 2.5 times more potentially harmful sites and phishing attacks. That is, compared to the previously used model.

In February this year, Chrome expanded its history to include so-called trips. The function allows you to map the browsing direction. The pages in history are thus grouped according to topics (eg planning a holiday in Ĺ umava), with the proviso that this happens only in a few selected languages, among which Czech is missing. Chrome uses machine learning to identify the language of the page. The goal is to offer you content in your language, translating the site if necessary.

Click for larger imageClick for larger image

With the next version of Chrome, another machine learning model will premiere. For a change, it will estimate the probability that you will have interest in the notifications offered by the website. He will hide potentially unwanted requests. According to Google, its browser also evaluates the probability today, but it cannot do so locally. That will change from the next issue.

Machine learning in the near future decides on the form of the navigation bar. This probably only applies to the mobile browser variant, where space for buttons is limited. Chrome may show a different button next to the address bar. If it senses that you would like to share, it will show the share button, other times it will make voice search available, for example. The evaluation will take place in real time.

Sources: The Keyword blog



Source link

Leave a Reply

Your email address will not be published.