Lightweight Compression of Intermediate Neural Network Features for Collaborative Intelligence

Abstract

In collaborative intelligence applications, part of a deep neural network (DNN) is deployed on a lightweight device such as a mobile phone or edge device, and the remaining portion of the DNN is processed where more computing resources are available, such as in the cloud. This paper presents a novel lightweight compression technique designed specifically to quantize and compress the features output by the intermediate layer of a split DNN, without requiring any retraining of the network weights. Mathematical models for estimating the clipping and quantization error of leaky-ReLU and ReLU activations at this intermediate layer are used to compute optimal clipping ranges for coarse quantization. We also present a modified entropy-constrained design algorithm for quantizing clipped activations. When applied to popular object-detection and classification DNNs, we were able to compress the 32-bit floating point intermediate activations down to 0.6 to 0.8 bits, while keeping the loss in accuracy to less than 1%. When compared to HEVC, we found that the lightweight codec consistently provided better inference accuracy, by up to 1.3%. The performance and simplicity of this lightweight compression technique makes it an attractive option for coding an intermediate layer of a split neural network for edge/cloud applications.

Crop Guidance Photography Algorithm for Mobile Terminals

“…Model pruning [26] is a commonly used model optimization technique aimed at reducing the size and complexity of deep learning models to improve their storage efficiency, computational efficiency, and generalization ability. The basic idea of model pruning is to reduce model complexity by removing redundant connections, reducing the number of parameters, or decreasing the number of layers while maintaining the model's performance on training and test data.…”