Optimization Methods for Large-Scale Machine Learning
Методы оптимизации для крупномасштабного машинного обучения
2018-01-01
SCID: 54.1/cf8cjwjs
Discuss with AI
deep neural network traininglarge-scale machine learningsecond-order optimizationstochastic gradient methodstext classification
Figures from the paper
Abstract (AI)
Abstract. This paper provides a review and commentary on the past, present, and future of numerical optimization algorithms in the context of machine learning applications. Through case studies on text classification and the training of deep neural networks, we discuss how optimization problems arise in machine learning and what makes them challenging. A major theme of our study is that large-scale machine learning represents a distinctive setting in which the stochastic gradient (SG) method has traditionally played a central role while conventional gradient-based nonlinear optimization techniques typically falter. Based on this viewpoint, we present a comprehensive theory of a straightforward, yet versatile SG algorithm, discuss its practical behavior, and highlight opportunities for designing algorithms with improved performance. This leads to a discussion about the next generation of optimization methods for large-scale machine learning, including an investigation of two main streams of research on techniques that diminish noise in the stochastic directions and methods that make use of second-order derivative approximations.
Key Findings
1
Future optimization methods are organized around reducing noise in stochastic directions and exploiting approximations to second-order derivatives.
2
Large-scale machine learning poses distinctive optimization challenges, illustrated through text classification and deep neural network training case studies.
3
Stochastic gradient methods have traditionally been central to large-scale machine learning, whereas conventional gradient-based nonlinear optimization techniques often perform poorly in this setting.
4
The paper develops a comprehensive theory and practical analysis of a straightforward, versatile stochastic-gradient algorithm.
5
The review identifies opportunities to improve large-scale machine-learning optimization by combining theoretical understanding with noise-reduction and second-order techniques.
Research Object
large-scale machine learning applications, including text classification and deep neural network training
Research Subject
the behavior, challenges, theory, and performance improvement of numerical optimization algorithms, especially stochastic-gradient methods and second-order or variance-reduction techniques
Publication Details
Publication Date
2018-01-01
Journal
Publisher
ISSN
Cited by
3224
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest