AI 中文总结
该研究受生物元可塑性启发，提出无任务持续学习框架SynGAP，通过自适应梯度预条件化缓解灾难性遗忘，在多个基准上表现优于现有方法，且内存高效。
AI 中文摘要
生物智能通过互补学习系统（CLS）理论自然避免灾难性遗忘，该宏观整合过程由局部层面的突触元可塑性驱动，即单个突触依赖历史的连续神经调节。人工神经网络在非平稳环境中面临稳定性-可塑性两难问题，现有解决方案常需任务标签或产生巨大内存开销，与生物现实不符。我们将这种局部神经调节重新表述为优化驱动过程，提出SynGAP：突触几何自适应预条件化，这是一种基于自适应梯度预条件化的无任务持续学习框架。SynGAP不依赖显式事件触发，而是通过在连续数据流上维护Fisher信息矩阵的指数移动平均来模拟实时元可塑性；在优化步骤中，这些动态元可塑性状态被转换为有界乘性掩码，用于预条件化原始梯度，选择性衰减对关键历史参数的更新。实验评估显示，与现有基线相比，SynGAP缓解灾难性遗忘的能力更优：在Split CIFAR-100基准上，SynGAP的准确率较EWC++提升4倍，比经验重放（ER）高出近10%，且两种方法的遗忘指标均降低超10%；在CORe50基准上，SynGAP达到约68%，比优化器基线提升10%。通过将连续生物元可塑性数学形式化为稳定的基于梯度的正则化，SynGAP为边缘自适应智能提供了高鲁棒性且内存高效的解决方案。
英文摘要
Biological intelligence naturally prevents catastrophic forgetting through Complementary Learning Systems (CLS) theory, a macroscopic consolidation process driven at the local level by synaptic metaplasticity: the continuous, history-dependent neuromodulation of individual synapses. While artificial neural networks struggle with the stability-plasticity dilemma in non-stationary environments, existing solutions often require task labels or incur massive memory overhead, diverging from biological reality. Re-framing this localized neuromodulation as an optimization-driven process, we introduce $\textbf{SynGAP}$: $\textbf{Syn}$aptic $\textbf{G}$eometric $\textbf{A}$daptive $\textbf{P}$reconditioning. SynGAP is a task-free continual learning framework based on adaptive gradient preconditioning. Rather than relying on explicit episodic triggers, SynGAP simulates real-time metaplasticity by maintaining an exponential moving average of the Fisher Information Matrix over a continuous data stream. During the optimization step, these dynamic metaplastic states are translated into a bounded multiplicative mask that preconditions raw gradients, selectively attenuating updates to critical historical parameters. Empirical evaluations demonstrate SynGAP's superior ability to mitigate catastrophic forgetting compared to established baselines. On the Split CIFAR-100 benchmark, SynGAP delivers a $4\times$ increase in accuracy compared to EWC++ and outperforms Experience Replay (ER) by almost $10\%$, while reducing the forgetting measure by over $10\%$ against both methods. Furthermore, on the CORe50 benchmark, SynGAP achieves about $68\%$, a $10\%$ improvement over optimizer baselines. By mathematically formalizing continuous biological metaplasticity as stable gradient-based regularization, SynGAP offers a highly robust and memory-efficient solution for adaptive intelligence at the edge.
