Compression-aware Projection with Greedy Dimension Reduction for Convolutional Neural Network Activations

TL;DR

This paper introduces a compression-aware projection system with greedy layer-wise optimization to reduce CNN activation memory access significantly while maintaining accuracy, enabling more efficient deployment on resource-constrained devices.

Contribution

It proposes a learnable projection and greedy selection metric for better accuracy-memory trade-off in activation compression, improving over existing methods.

Findings

Reduces memory access by 2.91x to 5.97x with negligible accuracy loss.

Effective on MobileNetV2, ResNet18, and VGG16 architectures.

Enhances deployment feasibility of CNNs on edge devices.

Abstract

Convolutional neural networks (CNNs) achieve remarkable performance in a wide range of fields. However, intensive memory access of activations introduces considerable energy consumption, impeding deployment of CNNs on resourceconstrained edge devices. Existing works in activation compression propose to transform feature maps for higher compressibility, thus enabling dimension reduction. Nevertheless, in the case of aggressive dimension reduction, these methods lead to severe accuracy drop. To improve the trade-off between classification accuracy and compression ratio, we propose a compression-aware projection system, which employs a learnable projection to compensate for the reconstruction loss. In addition, a greedy selection metric is introduced to optimize the layer-wise compression ratio allocation by considering both accuracy and #bits reduction simultaneously. Our test results show that the proposed methods effectively reduce 2.91x~5.97x memory access with negligible accuracy drop on MobileNetV2/ResNet18/VGG16.