About the Centre
In June 2025, the Higher School of Economics became one of the winners of the selection for the third wave of research centres in the field of artificial intelligence. The Centre for Optimisation and Adaptation of Large Fundamental Models at HSE (AI Centre) will focus on creating new methods and tools to make the training, use, and adaptation of complex artificial intelligence models cheaper and more efficient.
In addition to HSE, the winners included the Skolkovo Institute of Science and Technology, MIPT, the V.P. Ivannikov Institute of System Programming, Innopolis University, ITMO University, and Lomonosov Moscow State University.
Goals
- Development of new architectures and approaches to reduce the cost of training and inference, as well as to enhance the efficiency of adapting large foundational models;
- Creation and implementation of secure AI agents in the fields of medicine, education, banking, and other sectors of the economy.
Tasks
- Conducting domain adaptation of individual large foundational models with limited available computational resources;
- Developing tools to enhance the performance of existing pre-trained models;
- Advancing and creating new architectures for large foundational models that reduce the cost of inference and fine-tuning.
Expected scientific breakthroughs in the areas of foresight on AI
Architectures, Machine Learning (ML) algorithms, optimisation and mathematics
Effective training, retraining and acceleration algorithms for fundamental and generative models
Fundamental and generative models
Specialised AI agents for domains: biology, medicine, genetics, science and education
Safety, trust and explainability
Algorithmic assurance and tools for protecting AI models from attacks and manipulations
Management, decision-making, agent/multi-agent systems
Effective generation of data, reasoning chains and personalised recommendations