Laboratory of AI Security and Ethics

The application of the AI driven computations to solve practical tasks requires that AI algorithms are to be transparent to inspection, predictable to those they govern and robust against manipulation. The key research areas are

• development of approach to the modularization of the neural network models to achieve interpretability and robustness of the computations as well as parallel computations.
  
• development of a mathematical framework for modeling of the adversarial attacks on AI models.
  
• elaboration of the privacy-aware AI models, i.e. models implementing mechanisms for personal data protection when processing them using AI algorithms.