The most popular video hosting daily checks millions of rollers for the presence of prohibited or copyright materials, passing the most interesting video in the recommended. Make it quickly and efficiently helps artificial intelligence, which is applied in almost all sections YouTube.
In the first quarter of 2019, Youtube automatic mode deleted 8.3 million video. Algorithms are not perfect, sometimes completely «clean» materials are removed, but in most cases the AI system justifies itself. Scenes violence, demonstration of dangerous actions, drugs — all this is successfully detected at the download stage.
Before applying machine learning, malicious videos were removed using user complaints. According to statistics, only 8% of harmful material was seized until the 10th viewing. After the introduction of AI in the protection of Youtube, 50% of the prohibited content does not reach even 10 views.
Users remain unhappy if the offensive videos are still gaining hundreds of thousands views, falling into the recommended. Brand companies often order advertisements in YouTube and do not want it to appear in the prohibited rollers that hurt the reputation of the advertiser.
External pressure factors make Youtube developers working hard on the improvement of artificial intelligence to search for prohibited content. For this purpose, machine learning is used, which helps to organize the work of the «Classifier of Bad Video» on the home page and in the «Next» section. With the help of Big Data, comments are analyzed by the comments of the audience, which help to identify unacceptable content or «Klikbeit» — video with a loud title, the content of which does not correspond to the name.
The «Next» section includes videos that are similar in content to the video viewed. The Avtov Playback feature is enabled by the first roller list, located in the «Next» section. This list is formed in two stages. The first is analyzed by the history of the user’s views. At the second stage, an independent ranking system is created, which assigns to each video assessment.
It is also necessary to take into account that thousands of videos are downloaded on YouTube every minute, which can get into the list marked «New». Without the processing in real time, with the help of artificial intelligence and technology Big Data, YouTube would simply not cope with the issuance and instead of the long-awaited new issue of «Amkala» at the 7-year-old boy in the busplaces would have appeared another music video or another uninteresting video.
A huge flow of data is an ideal training platform for AI. It turned out that the Chelylands common on YouTube can not only be interesting, but also useful for studying augmented reality. As a result of processing more than 2000 video with popular in 2016, the programmers received a good basis for studying the depth of field sharpness to improve VR technologies, the ultimate goal of which is the effect of full presence.
The terrible trend of the streaming of terrorist acts began several years ago and continues now. The arrows in Galle and the terrorist attack in New Zealand are far from the first such cases.
In the United States already demanded that on all platforms with the possibility of striming, it was developed by software, allowing to identify dangerous personalities on direct broadcasts before they cause damage to civilians. YouTube and Facebook with the help of AI are already working to remove terrorist content in assistance with the Ministry of Justice and law enforcement agencies.
For the development of the canal, your support is important to us, subscribe to the channel and put the huskies.