An Extreme Learning Machine Based on Artificial Immune System

Algorithm 3

Artificial immune system extreme learning machine.

Step 1 Initialization

Randomly generate the initial antibody population

where .

Then calculate the fitness of by Eq.(12),

and get .

Step 2 Clone Selection

while the stop criterion is not met with do

    Step 2.1 Clone Phase

    For each antibody clone N-1 antibody, the clone pool named where

    =

    Step 2.2 Mutation Phase

    For each clone antibody , where and

    where is computed by Eq.(10).

    Then compute the fitness of by Eq.(12).

    And get .

    Step 2.3 Substitution Phase

    For each antibody , compare and

    For

end while

The final antibody population is and with the smallest fitness is the best antibody . Then

is used to opitimize the weight.

Step 3 ELM

Calculate the hidden-layer output matrix with the set of input weights and hidden biases represented by the .

Compute the output weights matrix .