Do Neural Networks Generalize from Self-Averaging Sub-classifiers in the Same Way As Adaptive Boosting?
This foundational research addresses the black-box nature of neural networks for the ML/AI community, offering a novel connection to boosting theory.
The paper tackles the problem of explaining why neural networks generalize by showing they learn a series of boosted classifiers with self-averaging over interpolating sub-classifiers, and provides experimental and theoretical evidence that NNs with dropout exhibit similar behavior as in boosting.
In recent years, neural networks (NNs) have made giant leaps in a wide variety of domains. NNs are often referred to as black box algorithms due to how little we can explain their empirical success. Our foundational research seeks to explain why neural networks generalize. A recent advancement derived a mutual information measure for explaining the performance of deep NNs through a sequence of increasingly complex functions. We show deep NNs learn a series of boosted classifiers whose generalization is popularly attributed to self-averaging over an increasing number of interpolating sub-classifiers. To our knowledge, we are the first authors to establish the connection between generalization in boosted classifiers and generalization in deep NNs. Our experimental evidence and theoretical analysis suggest NNs trained with dropout exhibit similar self-averaging behavior over interpolating sub-classifiers as cited in popular explanations for the post-interpolation generalization phenomenon in boosting.