Compressing Neural Networks: Towards Determining the Optimal Layer-wise Decomposition
This work addresses the challenge of efficiently compressing neural networks for deployment in resource-constrained environments, representing an incremental improvement over existing compression techniques.
The authors tackled the problem of compressing deep neural networks by developing a global compression framework that automatically determines optimal per-layer compression ratios while maintaining overall compression targets, achieving superior performance over existing low-rank compression methods across various networks and datasets.
We present a novel global compression framework for deep neural networks that automatically analyzes each layer to identify the optimal per-layer compression ratio, while simultaneously achieving the desired overall compression. Our algorithm hinges on the idea of compressing each convolutional (or fully-connected) layer by slicing its channels into multiple groups and decomposing each group via low-rank decomposition. At the core of our algorithm is the derivation of layer-wise error bounds from the Eckart Young Mirsky theorem. We then leverage these bounds to frame the compression problem as an optimization problem where we wish to minimize the maximum compression error across layers and propose an efficient algorithm towards a solution. Our experiments indicate that our method outperforms existing low-rank compression approaches across a wide range of networks and data sets. We believe that our results open up new avenues for future research into the global performance-size trade-offs of modern neural networks. Our code is available at https://github.com/lucaslie/torchprune.