Tag
This paper introduces CMP (Cognitive Memory Primitive), a continual-learning architecture that uses sparse relational codes and local learning to reduce catastrophic forgetting, demonstrating better backward transfer than a Transformer with EWC on a byte-level language modeling protocol.
This paper introduces Fisher width and inverse-Fisher width on statistical manifolds, studying their roles in local learning bounds and anisotropic recovery. It proves a complementary relation between the two widths and obtains recovery estimates based on Fisher geometry.
本文证明现代深度网络中预测编码（PC）使用的Jacobian转置乘积可以分解为局部可用的项，从而消除了自动求导反向传播的需要。提出的WF-Act-PC方法在CIFAR-10/100和Tiny-ImageNet上匹配甚至超越经过调优的反向传播基线，且性能随深度增加而提升。
This paper shows that predictive coding networks compute the same gradients as backpropagation in the limit of width much larger than depth, bridging biological learning and standard neural network training.
Proposes Adaptive Multi-Scale Goodness Aggregation (AMSGA), an extension of the Forward-Forward algorithm that improves stability, robustness, and generalization via multi-scale goodness aggregation, adaptive hard negative mining, and layer-dependent thresholds, achieving modest accuracy gains on MNIST and Fashion-MNIST.