Computing Gradient Vector and Jacobian Matrix in Arbitrarily Connected Neural Networks
- 19 August 2008
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 55 (10) , 3784-3790
- https://doi.org/10.1109/tie.2008.2003319
Abstract
This paper describes a new algorithm with neuron-by-neuron computation methods for the gradient vector and the Jacobian matrix. The algorithm can handle networks with arbitrarily connected neurons. The training speed is comparable with the Levenberg-Marquardt algorithm, which is currently considered by many as the fastest algorithm for neural network training. More importantly, it is shown that the computation of the Jacobian, which is required for second-order algorithms, has a similar computation complexity as the computation of the gradient for first-order learning methods. This new algorithm is implemented in the newly developed software, Neural Network Trainer, which has unique capabilities of handling arbitrarily connected networks. These networks with connections across layers can be more efficient than commonly used multilayer perceptron networks.Keywords
This publication has 28 references indexed in Scilit:
- Sensorless Control of Single Switch-Based Switched Reluctance Motor Drive Using Neural NetworkIEEE Transactions on Industrial Electronics, 2008
- Neural Network Based Method for Predicting Nonlinear Load HarmonicsIEEE Transactions on Power Electronics, 2007
- Robust Smooth-Trajectory Control of Nonlinear Servo Systems Based on Neural NetworksIEEE Transactions on Industrial Electronics, 2007
- Implementation of Artificial Neural Network-Based Tracking Controller for High-Performance Stepper Motor DrivesIEEE Transactions on Industrial Electronics, 2007
- A New TLS-Based MRAS Speed Estimation With Adaptive Integration for High-Performance Induction Machine DrivesIEEE Transactions on Industry Applications, 2004
- Fuzzy system based maximum power point tracking for PV systemPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Stator winding turn-fault detection for closed-loop induction motor drivesIEEE Transactions on Industry Applications, 2003
- Application of associative memory neural networks to the control of a switched reluctance motorPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- A stator-flux-oriented vector-controlled induction motor drive with space-vector PWM and flux-vector synthesis by neural networksIEEE Transactions on Industry Applications, 2001
- Simulation of neural networks to sensorless control of switched reluctance motorPublished by Institution of Engineering and Technology (IET) ,1998