Neural Epitome Search for Architecture-Agnostic Network Compression
This work addresses the limitation of fixed sampling strategies in model compression, offering a more flexible and efficient approach for compressing CNNs, though it is incremental as it builds upon existing methods like WSNet.
The paper tackles the problem of handcrafted weight sampling in network compression by proposing an auto-sampling method that learns sampling rules end-to-end, achieving a 6.5% improvement over WSNet on 1D convolution and outperforming MobileNetV2 by 1.47% accuracy with 25% FLOPs reduction on ImageNet.
The recent WSNet [1] is a new model compression method through sampling filterweights from a compact set and has demonstrated to be effective for 1D convolutionneural networks (CNNs). However, the weights sampling strategy of WSNet ishandcrafted and fixed which may severely limit the expression ability of the resultedCNNs and weaken its compression ability. In this work, we present a novel auto-sampling method that is applicable to both 1D and 2D CNNs with significantperformance improvement over WSNet. Specifically, our proposed auto-samplingmethod learns the sampling rules end-to-end instead of being independent of thenetwork architecture design. With such differentiable weight sampling rule learning,the sampling stride and channel selection from the compact set are optimized toachieve better trade-off between model compression rate and performance. Wedemonstrate that at the same compression ratio, our method outperforms WSNetby6.5% on 1D convolution. Moreover, on ImageNet, our method outperformsMobileNetV2 full model by1.47%in classification accuracy with25%FLOPsreduction. With the same backbone architecture as baseline models, our methodeven outperforms some neural architecture search (NAS) based methods such asAMC [2] and MNasNet [3].