arXiv Analytics

Sign in

arXiv:1703.00168 [stat.ML]AbstractReferencesReviewsResources

Modular Representation of Layered Neural Networks

Chihiro Watanabe, Kaoru Hiramatsu, Kunio Kashino

Published 2017-03-01Version 1

Deep neural networks have greatly improved the performance of various applications including image processing, speech recognition, natural language processing, and bioinformatics. However, it is still difficult to discover or interpret knowledge from the inference provided by a deep neural network, since its internal representation has many nonlinear and complex parameters embedded in hierarchical layers. Therefore, it becomes important to establish a new methodology by which deep neural networks can be understood. In this paper, we propose a new method for extracting a global and simplified structure from a layered neural network. Based on network analysis, the proposed method detects communities or clusters of units with similar connection patterns. We show its effectiveness by applying it to three use cases. (1) Network decomposition: it can decompose a trained neural network into multiple small independent networks thus dividing the problem and reducing the computation time. (2) Training assessment: the appropriateness of a trained result with a given hyperparameter or randomly chosen initial parameters can be evaluated by using a modularity index. And (3) data analysis: in practical data it reveals the community structure in the input, hidden, and output layers, which serves as a clue for discovering knowledge from a trained neural network.

Related articles: Most relevant | Search more
arXiv:1903.09215 [stat.ML] (Published 2019-03-21)
Empirical confidence estimates for classification by deep neural networks
arXiv:1804.04778 [stat.ML] (Published 2018-04-13)
Understanding Community Structure in Layered Neural Networks
arXiv:2310.01683 [stat.ML] (Published 2023-10-02)
Commutative Width and Depth Scaling in Deep Neural Networks