On-line learning in a discrete state space

1997/05/26 by Wolfgang Kinzel, W. Kinzel, Kinzel, W. +3
Computer Science · Physics and Astronomy · #Computability, Logic, AI Algorithms #Condensed Matter (cond-mat) #FOS: Physical sciences #Machine Learning and Algorithms #Neural Networks and Applications #cond-mat

paper · pdf · doi:10.48550/arxiv.cond-mat/9705257

7 pages, 1 Figure, Latex, submitted to J.Phys.A

arxiv created 1997/05/26 · openalex publication_date 1997/05/26 · arxiv updated 2009/11/30 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

On-line learning of a rule given by an N-dimensional Ising perceptron, is considered for the case when the student is constrained to take values in a discrete state space of size L^N. For L=2 no on-line algorithm can achieve a finite overlap with the teacher in the thermodynamic limit. However, if L is on the order of √(N), Hebbian learning does achieve a finite overlap.

Citations

• Optimal generalization in perceptions

Related

• On-line learning with restricted training sets: An exactly solvable case
• On-line learning through simple perceptron with a margin
• A Simple Perceptron that Learns Non-Monotonic Rules
• Dynamics of Learning with Restricted Training Sets II: Tests and Applications
• Learning by on-line gradient descent