Extremal Optimization Combined with LM Gradient Search for MLP Network Learning

  title={Extremal Optimization Combined with LM Gradient Search for MLP Network Learning},
  author={Peng Chen and Yongzai Lu and Yu-Wang Chen},
  journal={Int. J. Comput. Intell. Syst.},
Gradient search based neural network training algorithm may suffer from local optimum, poor generalization and slow convergence. In this study, a novel Memetic Algorithm based hybrid method with the integration of “extremal optimization” and “Levenberg–Marquardt” is proposed to train multilayer perceptron (MLP) networks. Inheriting the advantages of the two approaches, the proposed “EO-LM” method can avoid local minima and improve MLP network learning performance in generalization capability… 

Figures and Tables from this paper

Memetic algorithms based real-time optimization for nonlinear model predictive control

  • Peng ChenYongzai Lu
  • Engineering, Computer Science
    Proceedings 2011 International Conference on System Science and Engineering
  • 2011
A novel hybrid solution with the integration of bio-inspired computational intelligence extremal optimization and deterministic sequential quadratic programming for numerical optimization that is able to solve nonlinear programming (NLP) problems effectively is proposed.


The experimental results show the EO-SQP method is capable of finding the optimal or near optimal solutions for nonlinear programming problems effectively and efficiently.

Automatic Design of Robust Optimal Controller for Interval Plants using Genetic Programming and Kharitonov Theorem

The structure and parameters of the robust optimal controller for interval plants are optimized by Genetic Programming and the Generalized KT related stability criteria are integrated into the solution to guarantee the stability of the closed-loop system.

Lossless Compression of Medical Images Using a Dual Level DPCM with Context Adaptive Switching Neural Network Predictor

Experiments on magnetic resonance (MR) images showed lower prediction error for the DL-DPCM compared to the GAP and the MED, which are used in benchmark algorithms CALIC and LOCO-I respectively.

Robust Feedback Linearization Control for Reference Tracking and Disturbance Rejection in Nonlinear Systems

The method proposed ensures that a robust linear controller, designed for the linearized system obtained using robust feedback linearization, will maintain the robustness properties when applied to the initial nonlinear system.

Robust feedback linearization of an isothermal continuous stirred tank reactor: H∞ mixed-sensitivity synthesis and DK-iteration approaches

This paper studies the problem of designing a robust controller for the nonlinear multi-input multi-output continuous stirred tank reactor (CSTR) in the presence of uncertainties in parameters of the

Nonlinear Time-Varying Control Based on Differential Geometry

The actual physical system is a time-varying parameters expressed by the time-varying nonlinear system. For example, a robot manipulator system that has load variations and follows provided

Method of Extended Input/Output Linearization for the Time-Varying Nonlinear System

In this thesis, the method of input-output linearization for the time-varying nonlinear system is discussed. To this end, the proposed linearization technique of existing research has been extended

The Robust Input/Output Linearization Control for a Time-Varying Nonlinear System with Disturbance

A robust input-output linearization for time-varying nonlinear system is proposed and the robustlinearization for disturbance is proposed.



A tabu search algorithm for the training of neural networks

An efficient TS algorithm employing different strategies to provide a balance between intensification and diversification is proposed for the training of NNs and is found to outperform other metaheuristic techniques in the majority of the test cases.

Extremal Optimization: Methods derived from Co-Evolution

A general-purpose method for finding high-quality solutions to hard optimization problems, inspired by self-organized critical models of co-evolution such as the Bak-Sneppen model, which successively eliminates extremely undesirable components of sub-optimal solutions, rather than "breeding" better components.

Training feedforward networks with the Marquardt algorithm

The Marquardt algorithm for nonlinear least squares is presented and is incorporated into the backpropagation algorithm for training feedforward neural networks and is found to be much more efficient than either of the other techniques when the network contains no more than a few hundred weights.

Cellular Neural Networks-Based Genetic Algorithm for Optimizing the Behavior of an Unstructured Robot

It is demonstrated that due to the new method presented in this paper an irregular and even a disjointed walker robot can successfully move with the highest performance.

Solving Problems with Hidden Dynamics – Comparison of Extremal Optimisation and Ant Colony System

A performance comparison of Extremal optimisation and ant colony system is presented and it is endeavour to highlight particular strengths and weaknesses observed with different types of dynamic problem changes.

Efficient Initial Solution to Extremal Optimization Algorithm for Weighted MAXSAT Problem

An algorithm based on another distribution, known as the Bose-Einstein distribution in quantum physics, which provides a new stochastic initialization scheme to an Extremal Optimization procedure and is proposed for the approximated solution to an instance of the weighted maximum satisfiability problem (MAXSAT).