Abstract:Spiking neural network (SNN) is a brain-inspired model with more spatio-temporal information processing capacity and computational energy efficiency. However, with the increasing depth of SNNs, the memory problem caused by the weights of SNNs has gradually attracted attention. In this study, we propose an ultra-low latency adaptive local binary spiking neural network (ALBSNN) with accuracy loss estimators, which dynamically selects the network layers to be binarized to ensure a balance between quantization degree and classification accuracy by evaluating the error caused by the binarized weights during the network learning process. At the same time, to accelerate the training speed of the network, the global average pooling (GAP) layer is introduced to replace the fully connected layers by combining convolution and pooling. Finally, to further reduce the error caused by the binary weight, we propose binary weight optimization (BWO), which updates the overall weight by directly adjusting the binary weight. This method further reduces the loss of the network that reaches the training bottleneck. The combination of the above methods balances the network's quantization and recognition ability, enabling the network to maintain the recognition capability equivalent to the full precision network and reduce the storage space by more than 20%. So, SNNs can use a small number of time steps to obtain better recognition accuracy. In the extreme case of using only a one-time step, we still can achieve 93.39, 92.12, and 69.55% testing accuracy on three traditional static datasets, Fashion- MNIST, CIFAR-10, and CIFAR-100, respectively. At the same time, we evaluate our method on neuromorphic N-MNIST, CIFAR10-DVS, and IBM DVS128 Gesture datasets and achieve advanced accuracy in SNN with binary weights. Our network has greater advantages in terms of storage resources and training time.

A NOVEL STOCHASTIC BINARY NEURAL NETWORK

Binary stochasticity enabled highly efficient neuromorphic deep learning achieves better-than-software accuracy

Deep Spiking Neural Networks with Binary Weights for Object Recognition

ALBSNN: ultra-low latency adaptive local binary spiking neural network with accuracy loss estimator

STBNN: Hardware-friendly Spatio-Temporal Binary Neural Network with High Pattern Recognition Accuracy

Projected Stochastic Gradient Descent with Quantum Annealed Binary Gradients

High Performance and Hardware Efficient Stochastic Computing Elements for Deep Neural Network

sBSNN: Stochastic-Bits Enabled Binary Spiking Neural Network with On-Chip Learning for Energy Efficient Neuromorphic Computing at the Edge

An Energy-Efficient and Noise-Tolerant Recurrent Neural Network Using Stochastic Computing

An Energy-Efficient Bayesian Neural Network Implementation Using Stochastic Computing Method

Learning Adaptive Gradients for Binary Neural Networks

Quantized STDP-based online-learning spiking neural network

Novel Classification Algorithms Inspired by Firing Rate Stochastic Resonance

AdaBin: Improving Binary Neural Networks with Adaptive Binary Sets

BSNN: Towards faster and better conversion of artificial neural networks to spiking neural networks with bistable neurons

Ultra-low Latency Adaptive Local Binary Spiking Neural Network with Accuracy Loss Estimator

SA-BNN: State-Aware Binary Neural Network.

A Novel Conversion Method for Spiking Neural Network Using Median Quantization

Accelerating Stochastic Computation for Binary Classification Applications

Multi-Bit Mechanism: A Novel Information Transmission Paradigm for Spiking Neural Networks

Progressive Stochastic Binarization of Deep Networks