EDCompress: Energy-Aware Model Compression for Dataflows
This addresses the need for low-energy deployment of CNNs on edge devices, offering a novel method for optimizing across diverse hardware dataflows.
The paper tackles the problem of energy-aware model compression for CNNs on edge devices by proposing EDCompress, which improves energy efficiency by 20X, 17X, and 37X for VGG-16, MobileNet, and LeNet-5 networks with negligible accuracy loss.
Edge devices demand low energy consumption, cost and small form factor. To efficiently deploy convolutional neural network (CNN) models on edge device, energy-aware model compression becomes extremely important. However, existing work did not study this problem well because the lack of considering the diversity of dataflow types in hardware architectures. In this paper, we propose EDCompress, an Energy-aware model compression method for various Dataflows. It can effectively reduce the energy consumption of various edge devices, with different dataflow types. Considering the very nature of model compression procedures, we recast the optimization process to a multi-step problem, and solve it by reinforcement learning algorithms. Experiments show that EDCompress could improve 20X, 17X, 37X energy efficiency in VGG-16, MobileNet, LeNet-5 networks, respectively, with negligible loss of accuracy. EDCompress could also find the optimal dataflow type for specific neural networks in terms of energy consumption, which can guide the deployment of CNN models on hardware systems.