Compressing DNNs is important for the real-world applications operating on resource-constrained devices. However, we typically observe drastic performance deterioration when changing model size after training is completed. Therefore, retraining is required to resume the performance of the compressed models suitable for different devices. In this paper, we propose Decomposable-Net (the network decomposable in any size), which allows flexible changes to model size without retraining. We decompose weight matrices in the DNNs via singular value decomposition and adjust ranks according to the target model size. Unlike the existing low-rank compression methods that specialize the model to a fixed size, we propose a novel backpropagation scheme that jointly minimizes losses for both of full- and low-rank networks. This enables not only to maintain the performance of a full-rank network {\it without retraining} but also to improve low-rank networks in multiple sizes. Additionally, we introduce a simple criterion for rank selection that effectively suppresses approximation error. In experiments on the ImageNet classification task, Decomposable-Net yields superior accuracy in a wide range of model sizes. In particular, Decomposable-Net achieves the top-1 accuracy of $73.2\%$ with $0.27\times$MACs with ResNet-50, compared to Tucker decomposition ($67.4\% / 0.30\times$), Trained Rank Pruning ($70.6\% / 0.28\times$), and universally slimmable networks ($71.4\% / 0.26\times$).
翻译:压缩 DNNs 对在资源限制装置下操作的现实世界应用程序很重要。 但是, 我们通常观察到在培训完成后模型规模变化后, 性能会急剧恶化。 因此, 需要再培训才能恢复适合不同装置的压缩模型的性能。 在本文中, 我们提议以不再培训的方式灵活改变模型大小。 我们通过单值分解和根据目标模型大小调整级别, 在 DNS 中分解重量矩阵。 与目前将模型专门化为固定规模的低级压缩方法不同, 我们提出一个新的反向调整计划, 共同将全级和低级网络的损失降到最低。 这不仅能够保持全级网络的性能, 而能够改进多个规模的低级网络。 此外, 我们引入一个简单的级别选择标准, 有效抑制近似错误。 在图像网络分类任务中, 与现有的低级压缩压缩压缩压缩方法不同, 我们提出了新的反向后调整计划方案, 联合将全级和低级网络的损失降到最低级网络 。 具体地说, 与IMF AS- AS- AS- ASIM AS- AS- AS-0 ASIM AS- AS- AS- AS- AS- ASIM ASIM AS- AS- ASIM AS- AS-0 ASIM AS ASIM ASIM ASV AS AS AS ASIM ASIM AS AS ASV AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS ASIM ASIM ASVAS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS ASIM AS AS AS AS AS ASIM AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS AS