Prediction of evacuation time using extreme learning machine (ELM) neural network

Discussion in 'The Projects Forum' started by Bawal, Feb 3, 2015.

  1. Bawal

    Thread Starter New Member

    Feb 3, 2015
    Hello, I would like to have some guide on reducing the percentage error for this project. The project details are as below:
    Figure above is the structure of the chosen building for the evacuation time. 3 inputs are:
    a) Corridor width (1050 - 1400 mm)
    b) No. of room (4 -22)
    c) Lobby area (15 - 35 m2)

    750 samples data had been given in our project (unable to edit this part), where 500 of them is being used as training sample and 250 more for the test sample (cross validation).

    for ELM, weight and bias value can be randomly generated, and our professor only allow us to change the regularization ,C, and no. of hidden neuron,L for this project. we may add something incorporate with Artificial Intelligence subject or any engineering design.

    is there any method to determine the regularization and hidden neuron for lower percentage error on ELM?

    with here I attach the samples data together.

    thank you.