Iterative Low-Rank Approximation for CNN Compression
This work addresses the challenge of deploying large CNNs on resource-constrained embedded devices, presenting an incremental improvement in compression techniques.
The paper tackles the problem of compressing deep convolutional neural networks for embedded devices by proposing an iterative low-rank approximation method, achieving higher compression ratios with less accuracy loss compared to non-repetitive approaches.
Deep convolutional neural networks contain tens of millions of parameters, making them impossible to work efficiently on embedded devices. We propose iterative approach of applying low-rank approximation to compress deep convolutional neural networks. Since classification and object detection are the most favored tasks for embedded devices, we demonstrate the effectiveness of our approach by compressing AlexNet, VGG-16, YOLOv2 and Tiny YOLO networks. Our results show the superiority of the proposed method compared to non-repetitive ones. We demonstrate higher compression ratio providing less accuracy loss.