A Structurally Regularized CNN Architecture via Adaptive Subband Decomposition

TL;DR

This paper introduces a novel CNN architecture that decomposes input signals into subbands using adaptive filter banks, enhancing robustness, reducing computational costs, and achieving state-of-the-art accuracy on multiple image classification datasets.

Contribution

The proposed CNN architecture integrates adaptive subband decomposition into the learning process, providing structural regularization and computational efficiency while maintaining high accuracy.

Findings

Achieved over 90% reduction in inference computation cost.

Surpassed state-of-the-art accuracy on multiple datasets.

Demonstrated robustness to quantization noise.

Abstract

We propose a generalized convolutional neural network (CNN) architecture that first decomposes the input signal into subbands by an adaptive filter bank structure, and then uses convolutional layers to extract features from each subband independently. Fully connected layers finally combine the extracted features to perform classification. The proposed architecture restrains each of the subband CNNs from learning using the entire input signal spectrum, resulting in structural regularization. Our proposed CNN architecture is fully compatible with the end-to-end learning mechanism of typical CNN architectures and learns the subband decomposition from the input dataset. We show that the proposed CNN architecture has attractive properties, such as robustness to input and weight-and-bias quantization noise, compared to regular full-band CNN architectures. Importantly, the proposed architecture significantly reduces computational costs, while maintaining state-of-the-art classification accuracy. Experiments on image classification tasks using the MNIST, CIFAR-10/100, Caltech-101, and ImageNet-2012 datasets show that the proposed architecture allows accuracy surpassing state-of-the-art results. On the ImageNet-2012 dataset, we achieved top-5 and top-1 validation set accuracy of 86.91% and 69.73%, respectively. Notably, the proposed architecture offers over 90% reduction in computation cost in the inference path and approximately 75% reduction in back-propagation (per iteration) with just a single-layer subband decomposition. With a 2-layer subband decomposition, the computational gains are even more significant with comparable accuracy results to the single-layer decomposition.