Partial BFGS Update and Efficient Step-Length Calculation for Three-Layer Neural Networks

DOI10.1162/NECO.1997.9.1.123MaRDI QIDQ3123290zbMATH OpenOpenAlexDBLPWikidataFDO

Authors Kazumi Saito, Ryohei Nakano

Publication date 6 March 1997

Published in Neural Computation (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1162/neco.1997.9.1.123

zbMATH Keywords

quasi-Newton methodssecond-order learning algorithms

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Recommendations

• An efficient step size for BFGS
• Accelerating backpropagation using effective parameters at each step and an experimental evaluation
• Efficient calculation of the Gauss-Newton approximation of the Hessian matrix in neural networks
• Incremental approximation by layer neural networks
• scientific article; zbMATH DE number 1376915
• Advances in Neural Networks – ISNN 2005
• Extra multistep BFGS updates in quasi-Newton methods
• Sizing the BFGS and DFP updates: Numerical study
• BFGS with Update Skipping and Varying Memory

Cited in

(6)

• An empirical comparison of the validity of a neural net based multinomial logit choice model to alternative model specifications
• An adaptive Gauss-Newton algorithm for training multilayer nonlinear filters that have embedded memory
• Automated visual inspection system for wood defect classification using computational intelligence techniques
• Parametric Embedding for Class Visualization
• TDOR-MPINNs: multi-output physics-informed neural networks based on time differential order reduction for solving coupled Klein-Gordon-Zakharov systems
• Logistic regression using covariates obtained by product-unit neural network models

This page was built for publication: Partial BFGS Update and Efficient Step-Length Calculation for Three-Layer Neural Networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3123290)