International Journal of Man-Machine Studies - Special Issue: Knowledge Acquisition for Knowledge-based Systems. Part 5
Boolean Feature Discovery in Empirical Learning
Machine Learning
Learning approximate control rules of high utility
Proceedings of the seventh international conference (1990) on Machine learning
Issues in the design of operator composition systems
Proceedings of the seventh international conference (1990) on Machine learning
The general utility problem in machine learning
Proceedings of the seventh international conference (1990) on Machine learning
Learning internal representations by error propagation
Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Maintaining the utility of learned knowledge using model-based adaptive control
Maintaining the utility of learned knowledge using model-based adaptive control
Neural networks and the bias/variance dilemma
Neural Computation
Learning Search Control Knowledge: An Explanation-Based Approach
Learning Search Control Knowledge: An Explanation-Based Approach
Machine Learning
The effect of rule use on the utility of explanation-based learning
IJCAI'89 Proceedings of the 11th international joint conference on Artificial intelligence - Volume 1
IJCAI'89 Proceedings of the 11th international joint conference on Artificial intelligence - Volume 1
Concept learning and the problem of small disjuncts
IJCAI'89 Proceedings of the 11th international joint conference on Artificial intelligence - Volume 1
Concept formation over explanations and problem-solving experience
IJCAI'91 Proceedings of the 12th international joint conference on Artificial intelligence - Volume 2
Hi-index | 0.00 |
The overfit problem in inductive learning and the utility problem in speedup learning both describe a common behavior of machine learning methods: the eventual degradation of performance due to increasing amounts of learned knowledge. Plotting the performance of the changing knowledge during execution of a learning method (the performance response) reveals similar curves for several methods. The performance response generally indicates an increase to a single peak followed by a more gradual decrease in performance. The similarity in performance responses suggests a model relating performance to the amount of learned knowledge. This paper provides empirical evidence for the existence of a general model by plotting the performance responses of several learning programs. Formal models of the performance response are also discussed. These models can be used to control the amount of learning and avoid degradation of performance.