Efficient and Flexible Method for Reducing Moderate-size Deep Neural Networks with Condensation

TL;DR

This paper introduces a condensation-based reduction algorithm for moderate-size neural networks, effectively shrinking their size while preserving accuracy, thus enhancing inference speed for scientific and image classification tasks.

Contribution

The paper proposes a novel condensation reduction algorithm applicable to fully connected and convolutional networks, demonstrating significant size reduction with maintained performance.

Findings

Reduced network size to 41.7% in combustion tasks

Shrank CIFAR10 network to 11.5% of original size

Maintained prediction accuracy after reduction

Abstract

Neural networks have been extensively applied to a variety of tasks, achieving astounding results. Applying neural networks in the scientific field is an important research direction that is gaining increasing attention. In scientific applications, the scale of neural networks is generally moderate-size, mainly to ensure the speed of inference during application. Additionally, comparing neural networks to traditional algorithms in scientific applications is inevitable. These applications often require rapid computations, making the reduction of neural network sizes increasingly important. Existing work has found that the powerful capabilities of neural networks are primarily due to their non-linearity. Theoretical work has discovered that under strong non-linearity, neurons in the same layer tend to behave similarly, a phenomenon known as condensation. Condensation offers an opportunity to reduce the scale of neural networks to a smaller subnetwork with similar performance. In this article, we propose a condensation reduction algorithm to verify the feasibility of this idea in practical problems. Our reduction method can currently be applied to both fully connected networks and convolutional networks, achieving positive results. In complex combustion acceleration tasks, we reduced the size of the neural network to 41.7% of its original scale while maintaining prediction accuracy. In the CIFAR10 image classification task, we reduced the network size to 11.5% of the original scale, still maintaining a satisfactory validation accuracy. Our method can be applied to most trained neural networks, reducing computational pressure and improving inference speed.