BitSplit-Net: Multi-bit Deep Neural Network with Bitwise Activation Function
This work addresses efficiency challenges for deploying neural networks in hardware-limited settings, representing an incremental improvement over binary neural networks.
The paper tackles the problem of high computational cost and memory requirements in deep neural networks for resource-constrained environments by introducing BitSplit-Net, which uses multi-bit precision to improve accuracy while maintaining hardware-friendly characteristics, achieving similar classification accuracy at lower computational cost compared to conventional low-bit networks.
Significant computational cost and memory requirements for deep neural networks (DNNs) make it difficult to utilize DNNs in resource-constrained environments. Binary neural network (BNN), which uses binary weights and binary activations, has been gaining interests for its hardware-friendly characteristics and minimal resource requirement. However, BNN usually suffers from accuracy degradation. In this paper, we introduce "BitSplit-Net", a neural network which maintains the hardware-friendly characteristics of BNN while improving accuracy by using multi-bit precision. In BitSplit-Net, each bit of multi-bit activations propagates independently throughout the network before being merged at the end of the network. Thus, each bit path of the BitSplit-Net resembles BNN and hardware friendly features of BNN, such as bitwise binary activation function, are preserved in our scheme. We demonstrate that the BitSplit version of LeNet-5, VGG-9, AlexNet, and ResNet-18 can be trained to have similar classification accuracy at a lower computational cost compared to conventional multi-bit networks with low bit precision (<= 4-bit). We further evaluate BitSplit-Net on GPU with custom CUDA kernel, showing that BitSplit-Net can achieve better hardware performance in comparison to conventional multi-bit networks.