Prediction risk and architecture selection for neural networks

Prediction risk and architecture selection for neural networks (1994)

Cached

Download Links

by John Moody

Citations:

69 - 2 self

Versions

Version History

Metadata Version 1

Datum	Value	Source
TITLE	Prediction risk and architecture selection for neural networks	INFERENCE
AUTHOR NAME	John Moody	SVM HeaderParse 0.2
AUTHOR AFFIL	Oregon Graduate Institute, Department of Computer Science and Engineering, P.O. Box	SVM HeaderParse 0.2
AUTHOR ADDR	91000, Portland, OR 97291–1000,	SVM HeaderParse 0.2
ABSTRACT	Abstract. We describe two important sets of tools for neural network modeling: prediction risk estimation and network architecture selection. Prediction risk is defined as the expected performance of an estimator in predicting new observations. Estimated prediction risk can be used both for estimating the quality of model predictions and for model selection. Prediction risk estimation and model selection are especially important for problems with limited data. Techniques for estimating prediction risk include data resampling algorithms such as nonlinear cross–validation (NCV) and algebraic formulae such as the predicted squared error (PSE) and generalized prediction error (GPE). We show that exhaustive search over the space of network architectures is computationally infeasible even for networks of modest size. This motivates the use of heuristic strategies that dramatically reduce the search complexity. These strategies employ directed search algorithms, such as selecting the number of nodes via sequential network construction (SNC) and pruning inputs and weights via sensitivity based pruning (SBP) and optimal brain damage (OBD) respectively.	SVM HeaderParse 0.2
YEAR	1994	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	147--165	INFERENCE
CITATIONS	37 found	ParsCit 1.0