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

Semi-Supervised SRM code #108

Merged
merged 17 commits into from Sep 30, 2016

Conversation

Projects
None yet
3 participants
@TuKo
Contributor

TuKo commented Sep 20, 2016

Adding the Semi-Supervised SRM code, tests and examples.

In addition, I modified the raider dataset files in the examples for SRM and SS-SRM. The SRM examples (python and notebook) were corrected accordingly.

@TuKo

This comment has been minimized.

Show comment
Hide comment
@TuKo

TuKo Sep 20, 2016

Contributor

@cameronphchen Could you please review this PR as well? Thank you.

Contributor

TuKo commented Sep 20, 2016

@cameronphchen Could you please review this PR as well? Thank you.

@cameronphchen

This comment has been minimized.

Show comment
Hide comment
@cameronphchen

cameronphchen Sep 21, 2016

Contributor

@TuKo Sure, I'll do a review. Do you mind fixing the failed Travis CI first? I checked the error message, but not quite sure what's the cause.

Contributor

cameronphchen commented Sep 21, 2016

@TuKo Sure, I'll do a review. Do you mind fixing the failed Travis CI first? I checked the error message, but not quite sure what's the cause.

@TuKo

This comment has been minimized.

Show comment
Hide comment
@TuKo

TuKo Sep 21, 2016

Contributor

@cameronphchen, I looked at the build and didn't find any actual problem related to the code. It is failing only in one version of Mac and it is fine in the other one for Mac and Linux. The error does not seem related to the tests which is actually passing (https://travis-ci.org/IntelPNI/brainiak/jobs/161482100#L303). It is probably related to some library that we are using for the testing. I will check this with @mihaic later.

Please, go ahead and don't wait for this to be resolved. I will try to update during the day. Thanks.

Contributor

TuKo commented Sep 21, 2016

@cameronphchen, I looked at the build and didn't find any actual problem related to the code. It is failing only in one version of Mac and it is fine in the other one for Mac and Linux. The error does not seem related to the tests which is actually passing (https://travis-ci.org/IntelPNI/brainiak/jobs/161482100#L303). It is probably related to some library that we are using for the testing. I will check this with @mihaic later.

Please, go ahead and don't wait for this to be resolved. I will try to update during the day. Thanks.

@mihaic

This comment has been minimized.

Show comment
Hide comment
@mihaic

mihaic Sep 21, 2016

Contributor

@cameronphchen Indeed, the Travis error is not related to this PR. Please proceed with the review.

Contributor

mihaic commented Sep 21, 2016

@cameronphchen Indeed, the Travis error is not related to this PR. Please proceed with the review.

@mihaic

This comment has been minimized.

Show comment
Hide comment
@mihaic

mihaic Sep 28, 2016

Contributor

@cameronphchen Please remember about this review.

Contributor

mihaic commented Sep 28, 2016

@cameronphchen Please remember about this review.

@cameronphchen

This comment has been minimized.

Show comment
Hide comment
@cameronphchen

cameronphchen Sep 28, 2016

Contributor

Yes, I still remember this. Will try to finish before the Friday meeting.

On Wed, Sep 28, 2016, 13:29 Mihai Capotă notifications@github.com wrote:

@cameronphchen https://github.com/cameronphchen Please remember about
this review.


You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
#108 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ACqJEj81mj6IDeZw7i4S1mSS3omyjMMFks5quqQRgaJpZM4KCJUi
.

Contributor

cameronphchen commented Sep 28, 2016

Yes, I still remember this. Will try to finish before the Friday meeting.

On Wed, Sep 28, 2016, 13:29 Mihai Capotă notifications@github.com wrote:

@cameronphchen https://github.com/cameronphchen Please remember about
this review.


You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
#108 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ACqJEj81mj6IDeZw7i4S1mSS3omyjMMFks5quqQRgaJpZM4KCJUi
.

@cameronphchen

@TuKo @mihaic Overall looks good to me. Thank you Javier for sending in this PR!!! I have commented on some minor points.

Besides, is it possible to have some kind of test that's dependent on the dataset? For example, since we are using raider as the example dataset, can we have some kind of test that makes sure the results are always consistent on the raider dataset, e.g. image classification accuracy has to be X%+- 0.1%?

Show outdated Hide outdated brainiak/funcalign/sssrm.py
Show outdated Hide outdated brainiak/funcalign/sssrm.py
Show outdated Hide outdated brainiak/funcalign/sssrm.py
@@ -105,6 +105,71 @@ def fast_inv(a):
raise
def sumexp_stable(data):

This comment has been minimized.

@cameronphchen

cameronphchen Sep 29, 2016

Contributor

why do we need this instead of a softmax function directly? are we going to use this besides the calculation in softmax?

@cameronphchen

cameronphchen Sep 29, 2016

Contributor

why do we need this instead of a softmax function directly? are we going to use this besides the calculation in softmax?

This comment has been minimized.

@TuKo

TuKo Sep 30, 2016

Contributor

We are using this for the objective function computation. This is only for logging.

@TuKo

TuKo Sep 30, 2016

Contributor

We are using this for the objective function computation. This is only for logging.

@mihaic

This comment has been minimized.

Show comment
Hide comment
@mihaic

mihaic Sep 30, 2016

Contributor

@cameronphchen To test on the whole Raiders data we need long-running tests, which are different from the short-running tests we run now for PRs. We have an issue for documenting acceptable runtimes for the existing tests (issue #107) and an issue for long-running tests (issue #123). We can discuss those issues, but I think we should not hold this PR until they are solved; if you would like, you can create an SSSRM-specific testing issue blocked by issue #123.

Contributor

mihaic commented Sep 30, 2016

@cameronphchen To test on the whole Raiders data we need long-running tests, which are different from the short-running tests we run now for PRs. We have an issue for documenting acceptable runtimes for the existing tests (issue #107) and an issue for long-running tests (issue #123). We can discuss those issues, but I think we should not hold this PR until they are solved; if you would like, you can create an SSSRM-specific testing issue blocked by issue #123.

@cameronphchen

This comment has been minimized.

Show comment
Hide comment
@cameronphchen

cameronphchen Sep 30, 2016

Contributor

@mihaic Definitely, that shouldn't be blocking this PR.

Contributor

cameronphchen commented Sep 30, 2016

@mihaic Definitely, that shouldn't be blocking this PR.

@TuKo

This comment has been minimized.

Show comment
Hide comment
@TuKo

TuKo Sep 30, 2016

Contributor

Jenkins, retest this please.

Contributor

TuKo commented Sep 30, 2016

Jenkins, retest this please.

@TuKo

This comment has been minimized.

Show comment
Hide comment
@TuKo

TuKo Sep 30, 2016

Contributor

@cameronphchen , please review the latest changes. If everything is fine, please approve the PR. Otherwise let me know and I will add changes.
(Linux testing is failing because there is some issue with the system)

Contributor

TuKo commented Sep 30, 2016

@cameronphchen , please review the latest changes. If everything is fine, please approve the PR. Otherwise let me know and I will add changes.
(Linux testing is failing because there is some issue with the system)

@mihaic

This comment has been minimized.

Show comment
Hide comment
@mihaic

mihaic Sep 30, 2016

Contributor

Jenkins, retest this please.

Contributor

mihaic commented Sep 30, 2016

Jenkins, retest this please.

@cameronphchen

LGTM! Thanks!

@mihaic mihaic merged commit cbb3b98 into brainiak:master Sep 30, 2016

2 of 3 checks passed

continuous-integration/travis-ci/pr The Travis CI build is in progress
Details
linux Build finished.
Details
macos Build finished.
Details

@TuKo TuKo deleted the TuKo:semi3 branch Sep 30, 2016

danielsuo pushed a commit that referenced this pull request Nov 16, 2017

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