Laboratory for high dimension data
Grant Agreement No.: 14.Y26.31.0022
Project name: Scalable Artificial Intelligence Networks for Data Analysis in Growing Dimensions
Name of the institution of higher learning: National Research Lobachevsky State University of Nizhni Novgorod
Fields of scientific research: Computer and information sciences
In this project we will develop advanced methods for data mining in high dimension, optimised for growing dimensionality, including high dimensionality (dozens and hundreds) and very high dimensionality (thousands, tens thousands and higher). Such a technology is necessary in the world of fast growing data, explosive growth of production of AI systems, and fast development of non-stationary large heterogeneous networks of computers and gadgets. For this purpose we aim:
- To develop and explore new aspects and applications of the measure concentration theory (namely stochastic separation theorems for a wide families of data distributions and various classes of separation surfaces) as a background of the new technology.
- To develop methods for separation of genuinely high-dimensional problems from reducible problems that have low intrinsic dimension;
- To develop a theory, methodology and tools for creating new, and upgrading existing Big Data AI systems so that they are capable of learning on-the-fly from mistakes in real time.
- To develop theory and methods of dynamic models of optimal complexity based on the idea of “game against observer” (the worst case worlds);
- To adapt and apply the developed methods to the analysis of high dimensional data about biological neural nets (both in vitro and in viva).
- To generate solutions and open access software for high dimensional data analysis and for correction of large legacy AI systems.
- To generate specific and particular solutions for analysis of large and high dimensional live video streams, to complex biophysical, technical and hybrid man-machine systems.
- To create the Laboratory of advanced methods for high-dimensional data analysis and to provide its sustainable functioning.
Name: Gorban Alexanderч
Academic degree and title: Science Doctor (physics and mathematics)
Job Title: Professor, Chair in Applied Mathematics, Director of the Centre for Mathematical Modelling, University of Leicester
Field of scientific interests: Computer Science; Biophysics; Mathematical & Computational Biology; Mathematics; Physics; Thermodynamics; Chemical engineering; Interdisciplinary Sciences
2000 - Clay Scholar, Clay Mathematics Institute (Cambridge, USA)
2010 - Isaac Newton Institute for Mathematical Sciences Fellowship (Cambridge, UK)
1015 - Lifetime Achievement Award in recognition of outstanding contributions to the research field of (bio)chemical kinetics, International Conference MaCKIE-2015, Mathematics in (bio)Chemical Kinetics and Engineering, Ghent, Belgium
Scientific work of the leading scientist, his/her main scientific achievements:
Kinetics. A. Gorban has developed a family of methods for model reduction and coarse-graining: method of invariant manifold, method of natural projector, relaxation methods. He has solved problems in gas kinetics, polymer dynamics, chemical reaction kinetics and biological kinetics. For this series of work AG received the I. Prigogine award and medal, he has been a Clay Scholar (Cambridge, USA, 2000).
A. Gorban invented a new method to measure the stress caused by environmental factors. In particular, this is a possibility to measure the health of the groups of healthy people. This method is based on a universal effect discovered by A. Gorban in his study of human adaptation. This effect is supported by hundreds of experiments and observations and extended to systems of different nature. Now, this method is used for monitoring of Far North populations, for analysis of crises in national financial systems and in companies. It has become a part of the modern approach to crises anticipation.
A.S. Manso, M.H. Chai, J.M. Atack, L. Furi, M. De Ste Croix, R. Haigh, C. Trappetti, A.D. Ogunniyi, L.K. Shewell, M. Boitano, T.A. Clark, J. Korlach, M. Blades, E. Mirkes, A.N. Gorban, J.C. Paton, M.P. Jennings, M.R. Oggioni, A random six-phase switch regulates pneumococcal virulence via global epigenetic changes, 2014 Nature Communications
(Q1 IN ITERDISCIPLINARY SCIENCES)
A.N. Gorban, I. Karlin Hilbert's 6th Problem: exact and approximate hydrodynamic manifolds for kinetic equations 2014 Bulletin of the American Mathematical Society (Q1 IN MATHEMATICS)
F. Spahn, E. Vieira Neto, A.H.F. Guimarães, A.N. Gorban, N.V. Brilliantov A statistical model of aggregate fragmentation 2014 New Journal of Physics (Q1 IN PHYSICS, MULTIDISCIPLINARY)
A.N. Gorban, I. Tyukin, E. Steur, H. Nijmeijer Lyapunov-like conditions of forward invariance and boundedness for a class of unstable systems 2013 SIAM Journal On Control And Optimization (Q1 IN MATHEMATICS, APPLIED)
A.N. Gorban, G.S. Yablonsky Grasping Complexity 2013 Computers & Mathematics with
Applications (Q1 IN MATHEMATICS, APPLIED)