ALGORITHM OF THE MOST EFFICIENT USE OF MEMORY FOR A PAIRED MARKOV RANDOM FIELD

The article analyzes preliminary studies of a Markov random field. The energy function associated with a pairwise Markov random field is considered. An algorithm for the most efficient use of memory has been developed, which makes it possible to reduce the use of the occupied memory in comparison with standard algorithms for maximum loss. The features of its use are considered. The proposed algorithm converges in fewer iterations compared to the standard maximum flow image. Many foreign companies, where large amounts of data are collected, use machine learning and data mining techniques to process them. It is in most unstructured data that companies look for clues to solve a problem. But very often the tools used by companies do not allow to obtain reliable information. No less important is the speed of obtaining information. In order to improve the data processing process, machine learning algorithms are used. New algorithms for output in MRF are presented, which are more efficient in terms of operating time and / or memory usage) or efficient (in terms of solution quality) than the best modern methods. The calculations resulted in an algorithm for the most efficient use of memory, using fewer iterations, having a standard image of the maximum flow, without compromising optimality. The presented approach allows you to optimally solve larger problems on a standard computer. The presented algorithm establishes a method for selecting minimization of Isikawa-type graphs when a complete graph cannot be stored in memory.