Research map / Architectures and efficiency / Spiking networks

Efficient Training of Deep Spiking Neural Networks with Input-Driven Derivative-Free Updates

NeurIPS 2026 · Architectures and efficiency › Spiking networks · approach: Training deep spiking networks

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes an input-driven derivative-free training method for deep spiking neural networks as an alternative to BPTT with surrogate gradients, aiming to cut computation and memory costs.

Problem

Training spiking neural networks is hard because spike firing is non-differentiable; BPTT with surrogate gradients is effective but computationally and memory intensive.

Approach

Replaces surrogate-gradient backpropagation with input-driven derivative-free weight updates for deep SNNs, avoiding explicit gradient computation through spiking dynamics (details beyond the abstract not stated).

Experimental setup

Baselines: BPTT with surrogate gradients

Builds on: backpropagation through time surrogate gradient

Closest papers by meaning

• Temporal Efficient Training of Spiking Neural Network via Gradient Re-weighting (ICLR 2022 · similarity 0.91)
• Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.91)
• NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.91)
• Advancing Training Efficiency of Deep Spiking Neural Networks through Rate-based Backpropagation (NeurIPS 2024 · similarity 0.90)
• Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 2025 · similarity 0.90)
• Efficient Logit-based Knowledge Distillation of Deep Spiking Neural Networks for Full-Range Timestep Deployment (ICML 2025 · similarity 0.90)