📄 中文摘要
高维群体活动与局部可塑性机制的结合被认为是大脑中稳健感知的关键。然而，现代图像识别系统通常依赖于误差反向传播和端到端梯度优化，这与局部计算和局部可塑性并不自然对齐。RECAP（基于Hebbian共激活原型的水库计算）是一种受生物启发的学习策略，旨在通过将未训练的水库动态与自组织的Hebbian原型读出相结合，实现稳健的图像分类。该方法将时间平均的水库响应离散化为激活水平，构建水库单元对的共激活掩码，并逐步增强对重复结构的识别能力。
RECAP：作为水库动态自组织读出的局部Hebbian原型学习
📄 English Summary
RECAP: Local Hebbian Prototype Learning as a Self-Organizing Readout for Reservoir Dynamics
Robust perception in the brain is often attributed to the interplay of high-dimensional population activity and local plasticity mechanisms that reinforce recurring structures. In contrast, most contemporary image recognition systems rely on error backpropagation and end-to-end gradient optimization, which do not align naturally with local computation and plasticity. RECAP (Reservoir Computing with Hebbian Co-Activation Prototypes) is a bio-inspired learning strategy designed for robust image classification, integrating untrained reservoir dynamics with a self-organizing Hebbian prototype readout. The method discretizes time-averaged reservoir responses into activation levels, constructs a co-activation mask over pairs of reservoir units, and incrementally enhances the ability to recognize recurring structures.