Neurophysiology and Neural Network Modeling Group
The group's objective is to develop an algorithm for building an optimal neural network topology for pattern recognition and classification, clustering, decision-making, prediction, adaptive management, and various other tasks. We plan to achieve this by developing a model in which interacting neurons are treated as living cells.
Developing a system for effective neural network modeling based on available data, without manual adjustments for each specific task, will have a strong impact both on applied and research problems. For example, such modeling could be helpful in optimizing operating practices at manufacturing plants and improving performance indicators without changing technological process.
In the field of biology and medicine, such models, developed by our group, will help to study neurodegenerative processes leading to Alzheimer's and Parkinson's diseases. A model of schizophrenia development also exists, which is based on dopamine modulation in the prefrontal cortex. Modeling interactions in central nervous system also play an important role in experimental pharmacology and neurorehabilitation.
Global objectives of the group:
- Creating a molecular model of neural network functioning, which is as close to reality as possible.
- Developing a system to emulate evolutionary development of biologically approximate artificial neural network models (based on neurons functioning as living cells). Such system can be used to choose network models with optimal structure and external parameters for different tasks: pattern recognition, categorization, clustering, forecasting, decision-making, and adaptive management.
The project was initially launched in December of 2013, and since May of 2014 it has been operating as a part of JetBrains Research.