RePr: Improved Training of Convolutional Filters
This addresses the issue of inefficient filter training in CNNs for researchers and practitioners, offering a method to enhance performance without altering network architecture, though it is incremental as it builds on existing pruning techniques.
The paper tackles the problem of unnecessary overlap in convolutional filters by introducing a cyclic training method that temporarily prunes and restores filters, reducing feature overlap and improving generalization, with demonstrated performance gains across various tasks, especially for smaller networks.
A well-trained Convolutional Neural Network can easily be pruned without significant loss of performance. This is because of unnecessary overlap in the features captured by the network's filters. Innovations in network architecture such as skip/dense connections and Inception units have mitigated this problem to some extent, but these improvements come with increased computation and memory requirements at run-time. We attempt to address this problem from another angle - not by changing the network structure but by altering the training method. We show that by temporarily pruning and then restoring a subset of the model's filters, and repeating this process cyclically, overlap in the learned features is reduced, producing improved generalization. We show that the existing model-pruning criteria are not optimal for selecting filters to prune in this context and introduce inter-filter orthogonality as the ranking criteria to determine under-expressive filters. Our method is applicable both to vanilla convolutional networks and more complex modern architectures, and improves the performance across a variety of tasks, especially when applied to smaller networks.