Stochastic Downsampling for Cost-Adjustable Inference and Improved Regularization in Convolutional Networks

TL;DR

This paper introduces SDPoint, a stochastic downsampling method enabling CNNs to adapt their inference cost dynamically and improve regularization by training with random downsampling configurations.

Contribution

It proposes a novel stochastic downsampling technique for cost-adjustable inference and regularization in CNNs, allowing flexible inference budgets and enhanced model generalization.

Findings

SDPoint achieves effective cost-adjustable inference.

Sharing parameters across SDPoint instances provides regularization.

Extensive experiments validate improved performance and flexibility.

Abstract

It is desirable to train convolutional networks (CNNs) to run more efficiently during inference. In many cases however, the computational budget that the system has for inference cannot be known beforehand during training, or the inference budget is dependent on the changing real-time resource availability. Thus, it is inadequate to train just inference-efficient CNNs, whose inference costs are not adjustable and cannot adapt to varied inference budgets. We propose a novel approach for cost-adjustable inference in CNNs - Stochastic Downsampling Point (SDPoint). During training, SDPoint applies feature map downsampling to a random point in the layer hierarchy, with a random downsampling ratio. The different stochastic downsampling configurations known as SDPoint instances (of the same model) have computational costs different from each other, while being trained to minimize the same prediction loss. Sharing network parameters across different instances provides significant regularization boost. During inference, one may handpick a SDPoint instance that best fits the inference budget. The effectiveness of SDPoint, as both a cost-adjustable inference approach and a regularizer, is validated through extensive experiments on image classification.