Yandex has updated the search using neural networks: they also form responses to Alice

Get real time updates directly on you device, subscribe now.

Today, June 10, Yandex officially presented a major update to its search engine, which now uses generative neural networks YaLM for issuing – they are also used to compose texts in Russian and help the voice assistant “Alice” in forming responses to various queries. The Y1 update, according to the company, will allow the user to significantly save time searching for the necessary information – the system has learned to select the desired fragment on the video, based on a person’s request, or even give quick answers in a new issue format.

For example, as they said in Yandex, if a user enters the query “how to cook a tuna steak”, then the search engine will not only provide a video recipe, but also offer to start a video with tuna cooking exactly at the moment where they begin to tell the main essence of the recipe. To implement this technology, the company’s search engine compares the user’s request with the content of the videos – with an image and a sound track. Accordingly, the new issuance format will work with other instructions. Moreover, more quick answers will appear in the search results, which will become more diverse and useful.

If a user writes a request on how to open an IP, the search results will show a quick response and, best of all, will demonstrate related topics – who can open an IP, how much it costs, how long it takes, and so on. Yandex has also improved the smart camera technology, which is capable of recognizing objects, looking for where they are sold and how much they cost, as well as recognizing text and scanning documents. Now the camera works even better and is able to show information about objects in real time.

The innovations are already available on mobile devices, and the update should be released on the desktop version as soon as possible.

Get real time updates directly on you device, subscribe now.

Leave A Reply

Your email address will not be published.