The school of digital economy of DVFU creates the synthetic personality on the basis of artificial intelligence

On July 2, 2019 DVFU reported that they at its school of digital economy will create the digital body of Russian for training of machines, neuronets and development of the synthetic personality on the basis of artificial intelligence. The project is implemented in Laboratory of machine learning SHTSE DVFU based on the master educational program "Artificial intelligence and Big Data". Electronic collecting of applications for participation in the first stage of work will be open in September.

The digital manager — the synthetic personality on the basis of artificial intelligence, capable to support difficult dialogs with the user, to lead, find unevident answers and round the clock to solve service problems will become one of the first products according to the results of the carried-out work. On the similar principle languages can construct work of call centers, training systems, translators, different expert systems, management systems for difficult machine mechanisms.

We already began development of the synthetic personality in partnership with Sberbank, as led to setting of more global tasks. The lack of qualitatively marked base of Russian for training of neuronets became a serious call. We are going to answer it together with our technical partner in machine learning — Ashmanov's Neuronets company. It will provide us technology of a digital marking of material. We will transfer results of our joint work step by step for open use to all concerned parties, told Ilya Mirin, the principal of digital economy of DVFU

The expert explained that actually it is about preparation of the academic body of Russian which analogs on a global scale exist only for the English and French languages. The most important step on this way – to assemble the audio-body and to mark it in a special way, clear to the machine. Collecting of material will happen via the website and mobile application.

It is extremely volume work with perspective for many years. However we in ShTsE are going to complete primary stage of accumulation of language material in a year then we will start its digitization, explained Ilya Mirin, the principal of digital economy of DVFU

At the first stage volunteers from among students of DVFU will be involved. Further professional linguists, and specialists in computational linguistics who will be engaged in a qualitative marking of audiomaterial will be connected: will break it into word classes, will put down accents, pauses, will separate into dialogs and monologues, will result the said phrases in exact compliance to the written text, and the texts read at sight will separate from said naturally. At the same time it is necessary to solve the whole complex of satellite tasks.

Training data are not less important for development of algorithms of artificial intelligence, than algorithms. Emergence last decade of the open body from 14 million images of ImageNet had a great influence on development of computer vision – researchers and developers could create different methods of data analysis and apply computer vision in real tasks. Together with DVFU we will be able to collect "voice ImageNet" which will advance researches in the field of recognition and speech synthesis in Russia and the world. Besides, we will try to collect not only the Russian speech body, but also the body for languages of the small people of Russia, noted Stanislav Ashmanov, the CEO of Neuronets of Ashmanov

On a long interval of time those languages which had writing survived, and unwritten — practically died out. We speak about new writing — the language format suitable for training of machines. In this regard there was a danger that those languages in which machines — from microwaves and printers to cars and industrial robots will not talk — most likely, too will die out over time. For this reason language should be digitized, transferred it to model of a self-training neuronet. We will solve this important civilization problem in passing with development of applied products on the basis of artificial intelligence, summarized Ilya Mirin, the principal of digital economy of DVFU