Spatial synaptic regularization stabilizes learning across biological and artificial neural networks
Zhu Haoxing, Yinda Chen, Peng Zhao, Zhiwei Xiong, Hanchuan Peng, Feng Wu, Ruobing Zhang
This study identifies spatial synaptic organization as a mechanism for stabilizing learning and introduces Spatial Synaptic Regularization (SSR) to improve AI learning stability.
bioRxiv · Jun 30, 2026 · PreprintNeuroscience