A Review of Heuristic Global Optimization Based Artificial Neural Network Training Approahes

  title={A Review of Heuristic Global Optimization Based Artificial Neural Network Training Approahes},
  author={D. Geraldine Bessie Amali and M. Dinakaran},
  journal={IAES International Journal of Artificial Intelligence},
  • D. AmaliM. Dinakaran
  • Published 1 March 2017
  • Computer Science
  • IAES International Journal of Artificial Intelligence
Artificial Neural Networks have earned popularity in recent years because of their ability to approximate nonlinear functions. Training a neural network involves minimizing the mean square error between the target and network output. The error surface is nonconvex and highly multimodal. Finding the minimum of a multimodal function is a NP complete problem and cannot be solved completely. Thus application of heuristic global optimization algorithms that computes a good global minimum to neural… 

Figures and Tables from this paper

A New Quantum Tunneling Particle Swarm Optimization Algorithm for Training Feedforward Neural Networks

  • G. Amali DD. M
  • Computer Science
    International Journal of Intelligent Systems and Applications
  • 2018
A new Quantum Tunneling Particle Swarm Optimization (QTPSO) algorithm is proposed and applied to the training of feedforward Artificial Neural Networks (ANNs) and significantly outperforms three stateof-the-art PSO variants on a majority of benchmark neural network training problems.

Optimal unit commitment of a power plant using particle swarm optimization approach

It was found that, the minimum generation cost was achieved when all six generating units were committed and a total of 420,178.878$/h was saved.

A Review on Multi-Agent Technology in Micro-Grid Control

Current applications of MAS technology for MG both in basic and advanced control demands are reviewed, and a further improvement of performance is essential for future researches to increase penetration of MAS in MG control.

Identification of Web Risks in the Application layer using Machine Learning Techniques 2306

This paper is intended to look into the use of supervised-machine learning techniques which include; genetic algorithms and support-vector machines, as they are used in detecting some of the key web application layer threats.

Advances of metaheuristic algorithms in training neural networks for industrial applications

This article tries to compare and summarize the properties of various metaheuristic algorithms in terms of their convergence rate and the ability to avoid the local minima and categorizes the latest meta heuristic search algorithm in the literature to indicate their efficiency in training ANN for various industry applications.



A hybrid algorithm for artificial neural network training

Hybrid Artificial Bee Colony algorithm for neural network training

A hybrid algorithm combining Artificial Bee Colony (ABC) algorithm with Levenberq-Marquardt (LM) algorithm is introduced to train artificial neural networks (ANN). Training an ANN is an optimization

A Novel Hybrid Evolutionary Algorithm Based on PSO and AFSA for Feedforward Neural Network Training

FNN training by the novel hybrid evolutionary algorithm based on AFSA and PSO, also referred to as AFSA-PSO-parallel-hybrid evolutionary (APPHE) algorithm, has been used in FNN training and results are much more accurate and stable than by Levenberg-Marquardt back-propagation algorithm.

An Optimization Methodology for Neural Network Weights and Architectures

The approach combines the advantages of simulated annealing, tabu search and the backpropagation training algorithm in order to generate an automatic process for producing networks with high classification performance and low complexity.

A novel hybrid learning algorithm for full Bayesian approach of artificial neural networks

Design architectures and training of neural networks with a distributed genetic algorithm

Designing and training neural networks using a distributed genetic algorithm reinforced by the perceptron learning rule and examples are given showing the potential of the proposed approach.

Genetic algorithms and neural networks: optimizing connections and connectivity

A general CAC approach using novel ant algorithm training based neural network

  • Shenghong LiZemin Liu
  • Computer Science
    IJCNN'99. International Joint Conference on Neural Networks. Proceedings (Cat. No.99CH36339)
  • 1999
A neural network based approach for call admission control (CAC), which is applicable to very general traffic, and gives a novel ant algorithm to train the neural network.

Simulated annealing and weight decay in adaptive learning: the SARPROP algorithm

This work examines combining gradient descent with the global optimization technique of simulated annealing (SA), and shows the resulting algorithm, SARPROP, to be able to escape local minima, but is also able to maintain, and often improve the training times of the RPROP algorithm.

A SOFT-backpropagation algorithm for training neural networks

This work introduces a hybrid SOFT-BP algorithm where the SoFT-algorithm is applied first to obtain an initially good weight vector, which improves the precession of the weight vector to reach an acceptable error limit and shows an acceptable improvement in the training speed.