Congratulations to Petrosian Artur on the successful defense of his PhD thesis!

On January 10, 2023 a defence of the dissertation of Artur Petrosyan on "Modern methods of machine learning in the interpretation of electrical activity of the brain", presented for the degree of Candidate of Computer Sciences.
Supervisor: PhD Ossadtchi A.E.

Clinically relevant brain-computer interfaces (BCI) help rehabilitate patients with sensory, motor, and cognitive impairments. The task of decoding brain activity is one of the important components of BCI technology. The thesis research is divided into three parts. The first part describes a compact neural network architecture, taking into account generative models of brain electrical activity accepted in neurophysiology and physical principles of the registration process. Also the method of its interpretation is given and validated by means of simulations. In the second part, the developed methods were tested on a motor task. The advantages of the developed solution were demonstrated on real data. In the third part the developed methods were adapted to the task of speech decoding. Results of comparative analysis of quality of decoding words and parameters of the articulation path demonstrate the superiority of the offered architecture of a neural network in comparison with the similar solutions.

The text of the thesis, resume, and defense materials.