Abstract: The majority of the problems represented in the artificial intelligence is of combinatory nature and is characterized with exponential complexity. In the given lecture there are considered the meta-heuristics based methods of coping with dimensions of such problems. In particular, a combinatorial problem is considered as a sorting problem with constraints and is represented by means of a formalization of a searching of solutions in the state space. In the works we introduce a meta-heuristics of blocking, which allows a factorization of a state space and a reduction of an initial problem to a factor problem with a considerably smaller dimensions than it was an initial one. There is considered a mechanism of decomposition of an initial problem into the sub-problems and are represented conditions of correctness of merging sub-problems as well. Also in the lecture there is considered a usage of analogy principles in the process of solving combinatorial problems based on the blocking meta-heuristics.

Brief Biography of the Speaker:
Zurab Bosikashvili is a professor of Software Development and Artificial Intelligence at Information System Department, Georgian Technical University, Georgia. His area of expertise is the automatization of problem solving, pattern recognition, design of programming system and software development methodology. He authored or co-authored more 70 scientific papers published in reviewed journals or presented at local and international conferences. He has developed solutions searching methods and algorithms for combinatorial problems, particularly on their basis have been developed Georgian printed character and cursive script recognition system, logical blocks' control tests generation algorithm, system of conjunction tracing on the plane etc. He has participated more 30 projects in IT area of Georgia. He is a consultant and system architect in the software development company UGT.