|
Avtomatika i Telemekhanika, 2018, Issue 8, Pages 129–147
(Mi at14742)
|
|
|
|
This article is cited in 3 scientific papers (total in 3 papers)
Optimization, System Analysis, and Operations Research
Deep learning model selection of suboptimal complexity
O. Yu. Bakhteeva, V. V. Strijovb a Moscow Institute of Physics and Technology, Moscow, Russia
b Dorodnicyn Computing Centre, Russian Academy of Sciences, Moscow, Russia
Abstract:
We consider the problem of model selection for deep learning models of suboptimal complexity. The complexity of a model is understood as the minimum description length of the combination of the sample and the classification or regression model. Suboptimal complexity is understood as an approximate estimate of the minimum description length, obtained with Bayesian inference and variational methods. We introduce probabilistic assumptions about the distribution of parameters. Based on Bayesian inference, we propose the likelihood function of the model. To obtain an estimate for the likelihood, we apply variational methods with gradient optimization algorithms. We perform a computational experiment on several samples.
Keywords:
classification, regression, deep learning, model selection, Bayesian inference, variational inference, complexity.
Citation:
O. Yu. Bakhteev, V. V. Strijov, “Deep learning model selection of suboptimal complexity”, Avtomat. i Telemekh., 2018, no. 8, 129–147; Autom. Remote Control, 79:8 (2018), 1474–1488
Linking options:
https://www.mathnet.ru/eng/at14742 https://www.mathnet.ru/eng/at/y2018/i8/p129
|
Statistics & downloads: |
Abstract page: | 281 | Full-text PDF : | 84 | References: | 24 | First page: | 15 |
|