Particle Swarm Optimization Based Selective Ensemble of Online Sequential Extreme Learning Machine

Algorithm 2

PSOSEN.

Input: training set , learner , trial , threshold

Steps:

(1) for = 1 to

   = bootstrap sample from

   =

}

(2) generate a population of weight vectors

(3) evolve the population by PSO, where the fitness of the weight vector is defined as

.

(4) = the evolved best weight vector

Output: ensemble :

  for regression

  for classification