Improving Neural Network with Uniform Sparse Connectivity
This work provides a search-free, conceptually simple, and highly efficient sparse neural network architecture that can replace fully connected networks for a broad range of deep learning applications, offering substantial improvements for researchers and practitioners in AI.
This paper introduces the Uniform Sparse Network (USN) to address the high cost and overfitting issues of fully connected neural networks, as well as the complexity and suboptimal performance of existing sparse networks. USN achieves higher prediction accuracy than fully connected networks using only 0.55% of parameters and 25% of computing time, while also outperforming state-of-the-art sparse models in accuracy, speed, and robustness.
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.