Extreme learning machine based on length-changing particle swarm optimization algorithm

Abstract

Provided is an extreme learning machine based on a length-changing particle swarm optimization algorithm. The method comprises the steps that (1) the position and the speed of a particle swarm are initialized randomly, each particle in the particle swarm represents an ELM classifier; (2) an adaptation value f(p) for an evaluation function of each particle is computed; (3) the magnitude relationship of the row number of particles and the row number of globally optimal solution is compared, different updating formulas are selected to update the speed and the positions of the particles, a next generation of particle swarm is generated; (4) the optimum hidden unit number and corresponding input weight and hidden unit offset are selected; (5) output weight is computed, and the ELM classifier with the highest cross validation accuracy is obtained. The hidden unit number is automatically selected through the length-changing particle swarm optimization algorithm, and meanwhile the corresponding input weight and hidden unit offset are selected, so that the generalization performance of the ELM classifier is maximum, the ELM classifier with a small number of the hidden units can obtain the maximum generalization performance, testing time is short, and efficiency is high.

Description

technical field [0001] The invention relates to an extreme learning machine optimized and improved by using a variable-length particle swarm optimization algorithm, and belongs to the technical field of extreme learning machines. Background technique [0002] Single-hidden layer feedforward neural network (SLFN: Single-hidden Layer Feedforward Neural Network) can approximate any complex function with arbitrary precision, in 1998 in "IEEE Transactions on Neural Networks" (IEEE Neural Networks Journal), Volume 9, pp. 224-229 The published research on "The Upper Bound of the Number of Hidden Layer Neurons of Feedforward Neural Networks with Arbitrary Bounded Nonlinear Activation Functions" shows that a single hidden layer feedforward neural network with arbitrary nonlinear activation functions requires at most N hidden neurons. Layer neurons can learn N different samples with zero error, and have strong nonlinear identification capabilities. SLFNs have been widely used in patte...

AI Technical Summary

Problems solved by technology

The above strategy optimizes the number of ELM hidden elements, but does not optimize the input weight and hidden element bias at the same time.

[0010] In 2009, "Applied Soft Computing" (Applied Soft Computing) published "No-reference image quality assessment using modified extreme learning machine classifier" on pages 541-552 of volume 9, which proposed to apply the genetic algorithm of real number encoding to simultaneously optimize the number of hidden elements and corresponding Input weight and hidden element bias, but need to manually adjust multiple parameters of the new genetic operator defined in the algorithm

Images