New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

NEAT algorithm #752

Open
wants to merge 73 commits into
from

Conversation

Projects
None yet
4 participants
@BangLiu

BangLiu commented Aug 5, 2016

Implemented NEAT algorithm, and multiple test cases.

BangLiu added some commits May 17, 2016

CNE algorithm finished. XOR test passes. Need to revise coding style …
…and some TODOs. Then continue other algorithms.
seems bugs solved. keep printf. Seems information such as species siz…
…e, numSpecies are not updated when things changed, that is why so many bugs.
currently seems no bug. Previously we removed stale species even when…
… species number is small, leads to species disappear.
Removed redundant members in species, population. haven't remove in g…
…enome, as it caused some bug. Still need checking.
changed the place where we should set the childGenome's NumInput, Num…
…Output, GenomeDepth. However, seems calculate genome depth is quite slow.

BangLiu added some commits Jul 20, 2016

rebuild, solved compile problem. Add default constructors back. Witho…
…ut default constructor, there will have errors.
Revised neat.hpp. NeuronInnovation considers activation function type…
…. Revised MutateAddNode to be able to random select activation type.
Show outdated Hide outdated src/mlpack/methods/ne/neat.hpp
namespace mlpack {
namespace ne {
struct LinkInnovation {

This comment has been minimized.

@zoq

zoq Aug 18, 2016

Member

Can you add a comment that describes why this struct is useful.

@zoq

zoq Aug 18, 2016

Member

Can you add a comment that describes why this struct is useful.

This comment has been minimized.

Show outdated Hide outdated src/mlpack/methods/ne/neat.hpp
int newLinkInnovId;
};
struct NeuronInnovation {

This comment has been minimized.

@zoq

zoq Aug 18, 2016

Member

Can you add a comment that describes why this struct is useful.

@zoq

zoq Aug 18, 2016

Member

Can you add a comment that describes why this struct is useful.

This comment has been minimized.

@zoq

This comment has been minimized.

Show comment
Hide comment
@zoq

zoq Nov 8, 2016

Member

I've written a command line client and server to run the NEAT method, actually any mlpack classifier, against the OpenAI Gym toolkit. Which enables us to easily test the NEAT method against a variety of environments.

I've tried to reproduce the result for the Mario task, but it looks like the method doesn't converge as it did the last time I tested it. I'm not sure if the last change that effected the Evaluation process could be a reason for that. Maybe @BangLiu can provide some insights?

Member

zoq commented Nov 8, 2016

I've written a command line client and server to run the NEAT method, actually any mlpack classifier, against the OpenAI Gym toolkit. Which enables us to easily test the NEAT method against a variety of environments.

I've tried to reproduce the result for the Mario task, but it looks like the method doesn't converge as it did the last time I tested it. I'm not sure if the last change that effected the Evaluation process could be a reason for that. Maybe @BangLiu can provide some insights?

@BangLiu

This comment has been minimized.

Show comment
Hide comment
@BangLiu

BangLiu Dec 5, 2016

BangLiu commented Dec 5, 2016

@mlpack-jenkins

This comment has been minimized.

Show comment
Hide comment
@mlpack-jenkins

mlpack-jenkins May 24, 2017

Can one of the admins verify this patch?

Can one of the admins verify this patch?

@sourabhvarshney111

This comment has been minimized.

Show comment
Hide comment
@sourabhvarshney111

sourabhvarshney111 Mar 17, 2018

Contributor

@zoq I think this can also be closed because there is another PR for the same by @kartik-nighania.

Contributor

sourabhvarshney111 commented Mar 17, 2018

@zoq I think this can also be closed because there is another PR for the same by @kartik-nighania.

@zoq

This comment has been minimized.

Show comment
Hide comment
@zoq

zoq Mar 17, 2018

Member

I'd like to keep this open for some references, once NEAT is implemented we can close the PR.

Member

zoq commented Mar 17, 2018

I'd like to keep this open for some references, once NEAT is implemented we can close the PR.

@sourabhvarshney111

This comment has been minimized.

Show comment
Hide comment
@sourabhvarshney111

sourabhvarshney111 Mar 17, 2018

Contributor

Ok got it. Once the issue in the PR #1105 is solved, you would close this PR. Right @zoq.

Contributor

sourabhvarshney111 commented Mar 17, 2018

Ok got it. Once the issue in the PR #1105 is solved, you would close this PR. Right @zoq.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment