About Pyramid Structure in Convolutional Neural Networks

TL;DR

This paper explores the integration of pyramid structures in CNNs inspired by biological neurons, reducing parameters significantly while maintaining high accuracy across multiple datasets.

Contribution

It introduces a generalized framework for pyramid structures in CNNs that reduces parameters and disk size without sacrificing performance.

Findings

Over 80% parameter reduction in Caffe_LENET with maintained accuracy.

Achieved competitive results with 10-20% less training data and 10-40% fewer parameters in AlexNet.

Demonstrated effectiveness on MNIST, Cifar, and ImageNet datasets.

Abstract

Deep convolutional neural networks (CNN) brought revolution without any doubt to various challenging tasks, mainly in computer vision. However, their model designing still requires attention to reduce number of learnable parameters, with no meaningful reduction in performance. In this paper we investigate to what extend CNN may take advantage of pyramid structure typical of biological neurons. A generalized statement over convolutional layers from input till fully connected layer is introduced that helps further in understanding and designing a successful deep network. It reduces ambiguity, number of parameters, and their size on disk without degrading overall accuracy. Performance are shown on state-of-the-art models for MNIST, Cifar-10, Cifar-100, and ImageNet-12 datasets. Despite more than 80% reduction in parameters for Caffe_LENET, challenging results are obtained. Further, despite 10-20% reduction in training data along with 10-40% reduction in parameters for AlexNet model and its variations, competitive results are achieved when compared to similar well-engineered deeper architectures.