Towards Principled Design of Deep Convolutional Networks: Introducing SimpNet

TL;DR

This paper introduces SimpNet, a simple yet effective CNN architecture designed based on principled design guidelines, achieving a favorable accuracy-efficiency trade-off and outperforming complex models on several benchmarks.

Contribution

The paper proposes a set of fundamental design principles for efficient CNN architecture development and introduces SAF-pooling and SimpNet, a lightweight model that balances accuracy and computational cost.

Findings

SimpNet outperforms deeper architectures like VGGNet and ResNet on multiple benchmarks.

SimpNet achieves comparable or better accuracy with 2 to 25 times fewer parameters.

The approach demonstrates the effectiveness of principled design over ad-hoc methods.

Abstract

Major winning Convolutional Neural Networks (CNNs), such as VGGNet, ResNet, DenseNet, \etc, include tens to hundreds of millions of parameters, which impose considerable computation and memory overheads. This limits their practical usage in training and optimizing for real-world applications. On the contrary, light-weight architectures, such as SqueezeNet, are being proposed to address this issue. However, they mainly suffer from low accuracy, as they have compromised between the processing power and efficiency. These inefficiencies mostly stem from following an ad-hoc designing procedure. In this work, we discuss and propose several crucial design principles for an efficient architecture design and elaborate intuitions concerning different aspects of the design procedure. Furthermore, we introduce a new layer called {\it SAF-pooling} to improve the generalization power of the network while keeping it simple by choosing best features. Based on such principles, we propose a simple architecture called {\it SimpNet}. We empirically show that SimpNet provides a good trade-off between the computation/memory efficiency and the accuracy solely based on these primitive but crucial principles. SimpNet outperforms the deeper and more complex architectures such as VGGNet, ResNet, WideResidualNet \etc, on several well-known benchmarks, while having 2 to 25 times fewer number of parameters and operations. We obtain state-of-the-art results (in terms of a balance between the accuracy and the number of involved parameters) on standard datasets, such as CIFAR10, CIFAR100, MNIST and SVHN. The implementations are available at \href{url}{https://github.com/Coderx7/SimpNet}.