Hebbian学习与负反馈网络

简介：
《Hebbian学习与负反馈网络》一书探讨了神经网络中的Hebbian学习规则及其在构建高效负反馈系统中的应用，结合理论分析和实验验证，为人工智能领域提供了新思路。 Hebbian Learning and Negative Feedback Networks explores the concept that artificial neural networks, when utilizing negative feedback of activation, can employ simple Hebbian learning to self-organize in a way that uncovers significant patterns within datasets. The work considers two types of variants: one uses a single dataset for organization, with modifications to its learning rules demonstrating how it can execute Principal Component Analysis, Exploratory Projection Pursuit, Independent Component Analysis, Factor Analysis and various topology-preserving mappings. A second variant utilizes dual input data streams to self-organize, revealing in their basic form the capability of performing Canonical Correlation Analysis—a statistical method that identifies filters onto which projections from both datasets exhibit maximum correlation. The book delves into a broad spectrum of practical experiments and illustrates how these methodologies can be used to address real-world problems.