The need for small learning rates on large problems

Proceedings of the International Joint Conference on Neural Networks

Volumn 1, Issue , 2001, Pages 115-119

  Wilson, D R ^a   Martinez, T R ^a  

^a Fonix Corp   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

BACKPROPAGATION; COMPUTATIONAL COMPLEXITY; CONVERGENCE OF NUMERICAL METHODS; GRADIENT METHODS; LEARNING SYSTEMS; NEURAL NETWORKS; ONLINE SYSTEMS; PROBLEM SOLVING;

ERROR BACKPROPAGATION; LEARNING RATES; NEURAL NETWORK TRAINING; TRAINING ALGORITHM;

LEARNING ALGORITHMS;

EID: 0034856562     PISSN: None     EISSN: None     Source Type: Conference Proceeding    
DOI: None     Document Type: Conference Paper

References (8)

1. The inefficiency of batch training for large training sets
2. Increased rates of convergence through learning rate adaptation
3. SuperSAB: Fast adaptive back-propagation with good scaling properties
4. Accelerating the convergence of the back-propagation method