Maximizing Energy Efficiency in Spiking Neural Networks: A Dynamic Joint Pruning Framework

Spiking Neural Networks (SNNs) face increasing challenges for efficient deployment as architectures grow in complexity, necessitating network pruning to improve energy and computational efficiency. Existing pruning methods primarily focus on a single form of sparsity, overlooking the importance of joint pruning, which is critical for minimizing synaptic operations (SOPs) and enhancing energy efficiency. This paper presents a novel dynamic joint pruning framework that leverages both spatiotemporal spike sparsity and weight sparsity to minimize SOPs in SNN inference. Based on a comprehensive analysis of the SOPs model, we introduce an integrated solution that combines a multi-stage masking mechanism for fine-grained neuron firing threshold control, a temporal attention batch normalization (TABN) module with learnable time scaling factors, and a dynamic sparse strategy that adjusts importance coefficients based on real-time computational impact. Experimental results on CIFAR-10, CIFAR-100, and ImageNet validate the effectiveness of proposed framework. Our method achieves up to $126.38 \times$ compression ratio of SOPs on CIFAR-10 with minimal accuracy loss, establishing a new state-of-the-art in energy-efficient SNN pruning.