Due to recent updates, all users are required to create an Altair One account to login to the RapidMiner community. Click the Register button to create your account using the same email that you have previously used to login to the RapidMiner community. This will ensure that any previously created content will be synced to your Altair One account. Once you login, you will be asked to provide a username that identifies you to other Community users. Email us at Community with questions.

How can i use Differential Evolution to optimize parameters of neural network?

tassawar_alitassawar_ali Member Posts: 2 Learner II
edited June 2019 in Help
Anybody please guide me how can i use DE (advance form of GA) to optimize a neural net.
Tagged:

Answers

  • SGolbertSGolbert RapidMiner Certified Analyst, Member Posts: 344 Unicorn

    I didn't know the term, but I looked it up and it seems to be a variant of the generic evolutionary algorithm. So, to respond to the question, you can use the Optimize Parameters (Evolutionary) operator.

    To emulate the DE algorithm, you have to adapt the operator parameters a bit. In the DE algorithm the mutation of an agent is accepted only if it has a better fitness value. You could do something similar by generating N new agents in each iteration and then keeping the best N from the 2N agents that you have (+ selection). On the other hand, it is not always good to retain only the best agents, that's why the operator offers probabilistic selection methods.

    I have to confess that the operator's help is a bit lacking, there is no explanation of the selection types. For what I read, the rank selection method could be a good choice, as it tends to select the best agents but it also selects some of the worse performing from time to time.

    Regards,
    Sebastian

Sign In or Register to comment.