Energy-Aware Spike Budgeting for Continual Learning in Spiking Neural Networks for Neuromorphic Vision

2026-02-13 · Source: cs.NE updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Robotics & Autonomous Systems · Depth: Advanced, extended

Summary

A new energy-aware spike budgeting framework has been developed for continual learning in spiking neural networks (SNNs) to address catastrophic forgetting and optimize energy efficiency in neuromorphic vision systems. This framework integrates experience replay, learnable leaky integrate-and-fire (LIF) neuron parameters, and an adaptive spike scheduler. The approach exhibits modality-dependent behavior: on frame-based datasets like MNIST and CIFAR-10, it acts as a sparsity-inducing regularizer, improving accuracy while reducing spike rates by up to 47%. On event-based datasets such as DVS-Gesture, N-MNIST, and CIFAR-10-DVS, controlled budget relaxation enables accuracy gains up to 17.45 percentage points with minimal computational overhead. Evaluated across five benchmarks, the method consistently improves performance and minimizes dynamic power consumption, enhancing the practical viability of continual learning in energy-constrained neuromorphic vision.

Key takeaway

For Computer Vision Engineers developing neuromorphic systems, this framework offers a critical solution for deploying SNNs in continually evolving, energy-constrained environments. By adopting this energy-aware spike budgeting approach, you can achieve significant accuracy improvements and up to 47% spike rate reduction on frame-based data, or substantial accuracy gains on event-based data, without compromising energy efficiency. Consider implementing the adaptive spike scheduler and learnable LIF parameters to optimize your SNNs for diverse vision tasks.

Key insights

An energy-aware spike budgeting framework improves SNN continual learning by adaptively balancing accuracy and energy across vision modalities.

Principles

• Spike budgeting acts as a sparsity regularizer for frame-based data.
• Spike budgeting enables controlled activation for event-based data.
• Learnable neuron parameters enhance SNN adaptability.

Method

The framework uses a proportional control-based spike scheduler to dynamically adjust a regularization coefficient, weighting a penalty term in the loss function to enforce dataset-specific spike budgets.

In practice

• Apply adaptive spike scheduling to SNNs for energy optimization.
• Utilize learnable LIF parameters for improved SNN adaptability.
• Integrate experience replay to mitigate catastrophic forgetting.

Topics

• Spiking Neural Networks
• Continual Learning
• Neuromorphic Computing
• Event-Based Vision
• Energy Efficiency

Best for: Computer Vision Engineer, Research Scientist, AI Researcher, AI Scientist, AI Student

Related on AIssential

• Stable Spike: Dual Consistency Optimization via Bitwise AND Operations for Spiking Neural Networks — Dual consistency optimization via "AND" operations and amplitude-aware noise improves SNN performance and generalization.
• Low-Cost Neuromorphic Fall Detection Using Synthetic Event Data and Hybrid SNNs — Hybrid SNN-CNN models using synthetic event data offer energy-efficient, accurate fall detection for real-world applications.
• Spatio-Temporal Cluster-Triggered Encoding for Spiking Neural Networks — Cluster-based encoding preserves semantic structure in SNN spike trains, improving accuracy and sparsity.
• Event like spiking neuron lib that fits into the CPU cache [P] — CPU-cache-optimized spiking neuron library offers efficient data processing.
• End-to-End Radar and Communication Modulation Recognition with Neuromorphic Computing — EMRFormer leverages spiking neural networks and spike-driven transformers for energy-efficient automatic modulation recognition.
• Spike-Aware C++ INT8 Inference for Sparse Spiking Language Models on Commodity CPUs — Exploiting activation sparsity in spiking language models significantly boosts CPU inference throughput and reduces memory.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.NE updates on arXiv.org.