Alexey Ossadtchi gave a lecture at a seminar of the Artificial Intelligence Research Institute

On February 15, the director of the Center for Bioelectrical Interfaces, Alexey Ossadtchi, gave a lecture at a scientific seminar on artificial intelligence with the topic "Data analysis methods for studying brain activity in closed-loop paradigms.

Opponent: Sergey Shishkin, MSPU.

Modern neural network architectures take context into account by dynamically changing weight coefficients depending on input data. The addition of recurrent links allows to make the mappings realized by such models depend on the internal state of the latter. Similar principles are implemented in the human brain.

The standard methodology of neuroscientific research implies unidirectional interaction with the brain, in which subjects are presented with visual, auditory or tactile stimuli, the reaction is recorded, and then the variability of this reaction depending on the variation of parameters of such stimuli is studied. Recent closed-loop paradigms involve recurrent interaction with the brain, in which presented stimuli depend on the current state of the brain's neural networks.

Attempting to monitor the current state of the brain in real time and with low latency encounters fundamental and technical problems, the experience of overcoming which, as well as the applications of the technology under development, is the focus of this report.

The report is available here.