Learning Deep ResNet Blocks Sequentially using Boosting Theory
[edit]
Proceedings of the 35th International Conference on Machine Learning, PMLR 80:20582067, 2018.
Abstract
We prove a multichannel telescoping sum boosting theory for the ResNet architectures which simultaneously creates a new technique for boosting over features (in contrast with labels) and provides a new algorithm for ResNetstyle architectures. Our proposed training algorithm, BoostResNet, is particularly suitable in nondifferentiable architectures. Our method only requires the relatively inexpensive sequential training of $T$ “shallow ResNets”. We prove that the training error decays exponentially with the depth $T$ if the weak module classifiers that we train perform slightly better than some weak baseline. In other words, we propose a weak learning condition and prove a boosting theory for ResNet under the weak learning condition. A generalization error bound based on margin theory is proved and suggests that ResNet could be resistant to overfitting using a network with $l_1$ norm bounded weights.
Related Material


