An Efficient Compression of Deep Neural Network Checkpoints Based on Prediction and Context Modeling

This paper is dedicated to an efficient compression of weights and optimizer states (called checkpoints) obtained at different stages during a neural network training process. First, we propose a prediction-based compression approach, where values from the previously saved checkpoint are used for context modeling in arithmetic coding. Second, in order to enhance the compression performance, we also propose to apply pruning and quantization of the checkpoint values. Experimental results show that our approach achieves substantial bit size reduction, while enabling near-lossless training recovery from restored checkpoints, preserving the model's performance and making it suitable for storage-limited environments.

Distributed Video Coding Based on Polar Codes

In this letter we present an improved distributed video coding (DVC) scheme based on polar coding techniques. Firstly, we adapt log-likelihood ratios (LLRs) for DVC with integer implementation of a discrete cosine transform (DCT). We propose a computationally efficient and numerically stable modification of these LLRs based on the simplified methods of polar codes decoding. We show that on average this approach provides 0.3 dB PSNR gain for DVC with LDPC accumulated (LDPCA) codes. Secondly, we introduce the nested shortened polar codes construction algorithm. We demonstrate that replacement of LDPCA by polar codes improves PSNR by 0.1 dB on average, whereas, for videos with relatively high motion level, the gain reaches up to 0.23, 0.39 and 0.55 dB for Group of Pictures (GOP) lengths 2, 4 and 8 frames, respectively. Finally, experimental results demonstrate that DVC with polar codes and Tal-Vardy list decoder operates up to two times faster than DVC with LDPCA code and belief propagation (BP) decoder.