SEENN: Towards Temporal Spiking Early-Exit Neural Networks

2023/04/02 by Yuhang Li, Tamar Geller, Li, Yuhang +5 · 13 citations
Engineering · Neuroscience · #Advanced Memory and Neural Computing #FOS: Computer and information sciences #Ferroelectric and Negative Capacitance Devices #Machine Learning (cs.LG) #Neural and Evolutionary Computing (cs.NE) #Neural dynamics and brain function

paper · pdf · doi:10.48550/arxiv.2304.01230

openalex publication_date 2023/04/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Spiking Neural Networks (SNNs) have recently become more popular as a biologically plausible substitute for traditional Artificial Neural Networks (ANNs). SNNs are cost-efficient and deployment-friendly because they process input in both spatial and temporal manner using binary spikes. However, we observe that the information capacity in SNNs is affected by the number of timesteps, leading to an accuracy-efficiency tradeoff. In this work, we study a fine-grained adjustment of the number of timesteps in SNNs. Specifically, we treat the number of timesteps as a variable conditioned on different input samples to reduce redundant timesteps for certain data. We call our method Spiking Early-Exit Neural Networks (SEENNs). To determine the appropriate number of timesteps, we propose SEENN-I which uses a confidence score thresholding to filter out the uncertain predictions, and SEENN-II which determines the number of timesteps by reinforcement learning. Moreover, we demonstrate that SEENN is compatible with both the directly trained SNN and the ANN-SNN conversion. By dynamically adjusting the number of timesteps, our SEENN achieves a remarkable reduction in the average number of timesteps during inference. For example, our SEENN-II ResNet-19 can achieve 96.1% accuracy with an average of 1.08 timesteps on the CIFAR-10 test dataset. Code is shared at https://github.com/Intelligent-Computing-Lab-Yale/SEENN.

Cited by

• NeuRehab: A Reinforcement Learning and Spiking Neural Network-Based Rehab Automation Framework
• Hyperdimensional Decoding of Spiking Neural Networks
• A Closer Look at Knowledge Distillation in Spiking Neural Network Training
• ASTER: Attention-based Spiking Transformer Engine for Event-driven Reasoning
• A Scalable, Causal, and Energy Efficient Framework for Neural Decoding with Spiking Neural Networks
• Synergy Between the Strong and the Weak: Spiking Neural Networks are Inherently Self-Distillers
• Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training
• NeuCODEX: Edge-Cloud Co-Inference with Spike-Driven Compression and Dynamic Early-Exit
• SMTrack: End-to-End Trained Spiking Neural Networks for Multi-Object Tracking in RGB Videos
• STAS: Spatio-Temporal Adaptive Computation Time for Spiking Transformers
• ParaRevSNN: A Parallel Reversible Spiking Neural Network for Efficient Training and Inference
• Adaptive Time-step Training for Enhancing Spike-Based Neural Radiance Fields
• Edge Intelligence with Spiking Neural Networks

Related

• Direct Training Needs Regularisation: Anytime Optimal Inference Spiking Neural Network
• Unleashing the Potential of Spiking Neural Networks by Dynamic Confidence
• Knowing When to Stop: Delay-Adaptive Spiking Neural Network Classifiers with Reliability Guarantees
• SDSNN: A Single-Timestep Spiking Neural Network with Self-Dropping Neuron and Bayesian Optimization
• Improving Stability and Performance of Spiking Neural Networks through Enhancing Temporal Consistency