Recent advances in machine learning algorithms have determined outstanding breakthrough in many AI-driven application such as computer vision, speech recognition, autonomous vehicles, etc. However, there are open fundamental theoretical problems constraining world-wide application of the AI. Key focus areas include
- Elaboration of approaches to the construction of generalized neural network models, e.g. based on Sparse Factor Graph, that support cognitive structures and mechanisms such as attention, memory, context, belief, etc.
- Development of approaches to transition from computation models based on vectors to computation models based on sets.
- Accelerating neural network based computations by applying principles of tropical mathematics
- Development of methods of transition computations from independently and identically distributed features to out of distribution features
- Development of computer algebraic methods of multidimensional polynomial approximation.
- Development and analysis of quantum algorithms for artificial intelligence.