For the last seven years, he has been actively researching and developing computer vision and natural language processing systems. He is the author of a machine learning course on the Prometheus platform and an in-depth training course at the ARVI Lab.
He has extensive experience in video processing using deep learning methods for detecting objects and actions, predicting image depth maps, semantic segmentation and generating subtitles for images and video studios in Hollywood.
He has developed one of the first automation systems to control the placement of groceries at the store shelves using neural networks. He led the development of many projects for automated analysis of news in various languages, recognition of entities, analysis of conceptual drift and representation of language structures using machine learning systems.
During the last year, he worked on the system of automatic transfer of human facial features and body postures in the company NeoCortext, as well as on the task of the 4x resolution increasing for video studios in VideoGorillas.
Artificial neural networks allow automating classification, semantic segmentation, object recognition, etc.
Generative adversarial networks allow not only to handle a specific visual, audio, or text content but also to synthesize a new material that fits the features of “true” data.
This property can be used efficiently for the synthesis of personalized unique content “on the fly”.
In my opinion, such democratization of access to generative systems will radically change the online advertising market, content personalization, video, and photo production in the next few years.
We will talk about some examples of how generative neural networks already change some industries: the personal experience, the restoration of Orson Welles’s latest movie together with Netflix (which took two weeks instead of 6 months, due to these automatization methods). We will discuss the system of replacing face features for the personalization of advertising, memes, gif- animations and great enthusiasm from users, as well as examples of successful application of generative models for the synthesis of realistic text on a given topic.
We will discuss the moral and technological issues of such programs and proposals for countering the spread of high-quality fake news and methods for malware regulation.