Improving Neural Network with Uniform Sparse Connectivity

TL;DR

The paper introduces the uniform sparse network (USN), a simple, search-free neural network architecture that outperforms state-of-the-art sparse and fully connected models in accuracy, speed, and robustness, with significantly fewer parameters.

Contribution

USN is a novel, uniform sparse connectivity architecture that is independent of topology variation and model space, offering a simple, scalable alternative to fully connected networks.

Findings

USN outperforms state-of-the-art sparse networks in accuracy and speed.

USN achieves higher accuracy than fully connected networks with fewer parameters.

USN is open source and applicable across various architectures.

Abstract

Neural network forms the foundation of deep learning and numerous AI applications. Classical neural networks are fully connected, expensive to train and prone to overfitting. Sparse networks tend to have convoluted structure search, suboptimal performance and limited usage. We proposed the novel uniform sparse network (USN) with even and sparse connectivity within each layer. USN has one striking property that its performance is independent of the substantial topology variation and enormous model space, thus offers a search-free solution to all above mentioned issues of neural networks. USN consistently and substantially outperforms the state-of-the-art sparse network models in prediction accuracy, speed and robustness. It even achieves higher prediction accuracy than the fully connected network with only 0.55% parameters and 1/4 computing time and resources. Importantly, USN is conceptually simple as a natural generalization of fully connected network with multiple improvements in accuracy, robustness and scalability. USN can replace the latter in a range of applications, data types and deep learning architectures. We have made USN open source at https://github.com/datapplab/sparsenet.