DNNs as Layers of Cooperating Classifiers
This provides a novel perspective on generalization in DNNs, potentially advancing theoretical understanding for researchers, but it is incremental as it builds on existing work without broad empirical validation.
The paper tackles the lack of a robust theoretical framework for predicting generalization in deep neural networks by analyzing activation patterns, revealing that networks can be viewed as a combination of continuous and discrete information processing systems. It shows that these systems arise from gradient-based optimization and collaborate for classification, with their consistency being crucial for accuracy.
A robust theoretical framework that can describe and predict the generalization ability of deep neural networks (DNNs) in general circumstances remains elusive. Classical attempts have produced complexity metrics that rely heavily on global measures of compactness and capacity with little investigation into the effects of sub-component collaboration. We demonstrate intriguing regularities in the activation patterns of the hidden nodes within fully-connected feedforward networks. By tracing the origin of these patterns, we show how such networks can be viewed as the combination of two information processing systems: one continuous and one discrete. We describe how these two systems arise naturally from the gradient-based optimization process, and demonstrate the classification ability of the two systems, individually and in collaboration. This perspective on DNN classification offers a novel way to think about generalization, in which different subsets of the training data are used to train distinct classifiers; those classifiers are then combined to perform the classification task, and their consistency is crucial for accurate classification.