发表机构
Beijing Institute of Technology; Capital Medical University; Lanzhou University（北京理工大学; 首都医科大学; 兰州大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出梯度隧道算法，利用局部脉冲时序解决神经微电路的时间信用分配问题，在真实基准上以更少参数达到领先性能，并解释了大脑学习机制。
AI 中文摘要
大脑使用离散脉冲进行动态计算，然而，神经微电路（NMCs）如何利用局部脉冲时序解决时间信用分配问题仍是一个基本未解之谜。主流的脉冲神经网络（SNN）方法通过代理梯度近似反向传播来规避这一问题，从而将学习与生物脉冲时序脱钩。在此，我们将时间信用分配重新表述为一个状态分离问题：直接从当前神经状态中提取由历史扰动引发的任务所需成分。这通过梯度隧道（GT）算法和超前-滞后展开技术，为NMCs实现了一个在线反馈学习框架，该框架从局部突触脉冲时序中推导信用分配，同时与ANN-SNN混合架构兼容。实验上，GT训练的NMCs在长时间尺度证据整合和抗噪记忆保持方面表现出色，并在真实世界基准测试中以远少于其他方法的参数达到与领先SNN在线学习方法相当的性能。所提出的框架解决了长达二十年之久的NMC反馈学习问题，并为大脑的学习机制提供了一种计算上合理的解释。
英文摘要
The brain uses discrete spikes for dynamic computation, yet, how neural microcircuits (NMCs) solve temporal credit assignment using local spike timing remains a fundamental open question. Dominant spiking neural network (SNN) approaches circumvent this by approximating backpropagation through surrogate gradients, decoupling learning from biological spike timing. Here, we reformulate temporal credit assignment as a state separation problem: extracting task-required components induced by historical perturbations directly from the current neural state. This enables an online feedback learning framework for NMCs through a gradient tunneling (GT) algorithm and the lead-lag expansion technique that derives credit assignment from local synaptic spike timing, while remaining compatible with ANN-SNN hybrid architectures. Experimentally, GT-trained NMCs excel at long-timescale evidence integration and noise-robust memory retention, and perform comparably to leading SNN online learning methods on real-world benchmarks with far fewer parameters. The proposed framework addresses the two-decade-old NMC feedback learning problem and suggests a computationally plausible explanation for the brain's learning mechanisms.
CommentsVersion for the initial submission to Nature Machine Intelligence, 10th Sep. 2026
