Deep Roots: Improving CNN Efficiency with Hierarchical Filter Groups

TL;DR

This paper introduces a hierarchical filter group structure for CNNs that significantly reduces computational costs and parameters while maintaining or improving accuracy, validated on CIFAR10 and ILSVRC datasets.

Contribution

The authors propose a novel sparse, tree-like connection structure for CNNs that enhances efficiency without sacrificing accuracy.

Findings

ResNet 50: 40% fewer parameters, 45% fewer FLOPs, 31% faster on CPU

ResNet 200: 25% fewer FLOPs, 44% fewer parameters, same accuracy

GoogLeNet: 7% fewer parameters, 21% faster on CPU

Abstract

We propose a new method for creating computationally efficient and compact convolutional neural networks (CNNs) using a novel sparse connection structure that resembles a tree root. This allows a significant reduction in computational cost and number of parameters compared to state-of-the-art deep CNNs, without compromising accuracy, by exploiting the sparsity of inter-layer filter dependencies. We validate our approach by using it to train more efficient variants of state-of-the-art CNN architectures, evaluated on the CIFAR10 and ILSVRC datasets. Our results show similar or higher accuracy than the baseline architectures with much less computation, as measured by CPU and GPU timings. For example, for ResNet 50, our model has 40% fewer parameters, 45% fewer floating point operations, and is 31% (12%) faster on a CPU (GPU). For the deeper ResNet 200 our model has 25% fewer floating point operations and 44% fewer parameters, while maintaining state-of-the-art accuracy. For GoogLeNet, our model has 7% fewer parameters and is 21% (16%) faster on a CPU (GPU).