Abstract:The Spiking Neural Network (SNN), due to its unique spiking-driven nature, is a more energy-efficient and effective neural network compared to Artificial Neural Networks (ANNs). The encoding method directly influences the overall performance of the network, and currently, direct encoding is primarily used for directly trained SNNs. When working with static image datasets, direct encoding inputs the same feature map at every time step, failing to fully exploit the spatiotemporal properties of SNNs. While temporal encoding converts input data into spike trains with spatiotemporal characteristics, traditional SNNs utilize the same neurons when processing input data across different time steps, limiting their ability to integrate and utilize spatiotemporal information <a class="link-external link-http" href="http://effectively.To" rel="external noopener nofollow">this http URL</a> address this, this paper employs temporal encoding and proposes the Adaptive Spiking Neural Network (ASNN), enhancing the utilization of temporal encoding in conventional SNNs. Additionally, temporal encoding is less frequently used because short time steps can lead to significant loss of input data information, often necessitating a higher number of time steps in practical applications. However, training large SNNs with long time steps is challenging due to hardware constraints. To overcome this, this paper introduces a hybrid encoding approach that not only reduces the required time steps for training but also continues to improve the overall network performance.Notably, significant improvements in classification performance are observed on both Spikformer and Spiking ResNet architectures.our code is available at <a class="link-external link-https" href="https://github.com/hhx0320/ASNN" rel="external noopener nofollow">this https URL</a>

Training Spiking Neural Networks for Reinforcement Learning Tasks with Temporal Coding Method

A Low Latency Adaptive Coding Spike Framework for Deep Reinforcement Learning

Efficient Spiking Neural Networks with Logarithmic Temporal Coding.

Training Spiking Neural Networks for Cognitive Tasks: A Versatile Framework Compatible with Various Temporal Codes.

A Hybrid Learning Framework for Deep Spiking Neural Networks with One-Spike Temporal Coding

A Low Latency Adaptive Coding Spiking Framework for Deep Reinforcement Learning

An Efficient Learning Algorithm for Direct Training Deep Spiking Neural Networks

Efficient Spiking Neural Networks with Sparse Selective Activation for Continual Learning

A Brain-Inspired Spiking Neural Network Model with Temporal Encoding and Learning

Deep Spiking Neural Network Using Spatio-temporal Backpropagation with Variable Resistance.

Temporal Efficient Training of Spiking Neural Network via Gradient Re-weighting

Supervised Learning Based on Temporal Coding in Spiking Neural Networks

Spike Attention Coding for Spiking Neural Networks.

Temporal Neural Encoding Methods for Spiking Neural Networks

Adaptive Multi-Level Firing for Direct Training Deep Spiking Neural Networks

IM-LIF: Improved Neuronal Dynamics with Attention Mechanism for Direct Training Deep Spiking Neural Network

Temporal-Coded Deep Spiking Neural Network with Easy Training and Robust Performance.

Delay learning based on temporal coding in Spiking Neural Networks

Enabling Spike-Based Backpropagation for Training Deep Neural Network Architectures

Temporal-Coded Spiking Neural Networks with Dynamic Firing Threshold: Learning with Event-Driven Backpropagation

Adaptive Spiking Neural Networks with Hybrid Coding