|
MATH MODELING
Approximate estimation using the accelerated maximum entropy method. Part 1. Problem statement and implementation for the regression problem
Yu. A. Dubnovab, A. V. Boulytchevab a Federal Research Center "Computer Science and Control" of Russian Academy of Sciences, Moscow, Russia
b HSE University, Moscow
Abstract:
The work is devoted to the development of an entropy estimation method with “soft” randomization for restoring the parameters of probabilistic mathematical models from the available observations. Soft randomization refers to the technique of adding regularization to the information entropy functional to simplify the optimization problem and speed up learning process compared to the traditional maximum entropy method. In this work, the concept of the soft randomization entropy estimation method was developed, including obtaining entropy-optimal PDF functions in general form. During the experiments, several types of model regularization were tested on the example of a classical regression analysis problem.
Keywords:
probabilistic mathematical model, maximum entropy method, linear regression, regularization.
Citation:
Yu. A. Dubnov, A. V. Boulytchev, “Approximate estimation using the accelerated maximum entropy method. Part 1. Problem statement and implementation for the regression problem”, Informatsionnye Tekhnologii i Vychslitel'nye Sistemy, 2022, no. 4, 69–80
Linking options:
https://www.mathnet.ru/eng/itvs787 https://www.mathnet.ru/eng/itvs/y2022/i4/p69
|
Statistics & downloads: |
Abstract page: | 28 | Full-text PDF : | 6 | First page: | 2 |
|