[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user asfgit closed the pull request at: https://github.com/apache/spark/pull/3890 --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enab

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user marmbrus commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-137565250 What is the status here and why is this PR deleting UDF registration? Can we close this issue? --- If your project is set up for it, you can reply to this email and h

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98190811 I mean for svm the formulation is over all rows right...the smooth max will be done on every row and label...max(0, 1 - y_i a_i*x)...so only change will be a diff func

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dlwh commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98190256 ? On Fri, May 1, 2015 at 11:03 AM, Debasish Das wrote: > nope...logistic is feature space...svm is data space...the gradient > calculation / BFGS Co

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98189044 nope...logistic is feature space...svm is data space...the gradient calculation / BFGS CostFun will change --- If your project is set up for it, you can reply to

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dlwh commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98188805 isn't that just softmax/logistic regression then? On Fri, May 1, 2015 at 11:00 AM, Debasish Das wrote: > @dlwh we should simpl

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98188550 @dlwh we should simply use your smooth max and make max(0, 1 - ya'x) differentiable for the first version...that needs no change to breeze...and then if needed we use

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dbtsai commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98175466 @debasish83, after doing some research, we think subLBFGS https://users.soe.ucsc.edu/~niejiazhong/slides/vishy-lec4.pdf looks like the way to go. @devesh created a feature

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98073720 this is linear svm strictly in primal form...there are ways to fix it through going to dual space but that needs a linear / nonlinear kernel generation which might be

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-98073658 @loachli hinge loss in linear svm is max(0, 1 - y*a'x) right ? Just replace max with a smooth max and you should be able to smooth hinge gradient and then it can be di

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user loachli commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-86268363 @debasish83 , in the paper, owlqn is designed for logistic regression + L1. I do not know whether it is suitable for svm. Owlqn in breeze supports elasticnet that linea

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user debasish83 commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84624771 Can we discuss it in JIRA ? For svm with owlqn what's the orthant wise constraint you are adding ? There are ways to handle the max differentiability in bfgs as well b

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user AmplabJenkins commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84609099 Test FAILed. Refer to this link for build results (access rights to CI server needed): https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84609071 [Test build #28964 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/28964/consoleFull) for PR 3890 at commit [`202c34e`](https://gith

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84608248 [Test build #28964 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/28964/consoleFull) for PR 3890 at commit [`202c34e`](https://githu

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84599277 [Test build #28963 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/28963/consoleFull) for PR 3890 at commit [`be0db7c`](https://gith

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user AmplabJenkins commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84599278 Test FAILed. Refer to this link for build results (access rights to CI server needed): https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/28

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-84598782 [Test build #28963 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/28963/consoleFull) for PR 3890 at commit [`be0db7c`](https://githu

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user AmplabJenkins commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-70843039 Test PASSed. Refer to this link for build results (access rights to CI server needed): https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/25

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-70843031 [Test build #25889 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25889/consoleFull) for PR 3890 at commit [`020a9e2`](https://gith

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-70832740 [Test build #25889 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25889/consoleFull) for PR 3890 at commit [`020a9e2`](https://githu

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dbtsai commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-69624254 @loachli OWLQN doesn't automatically solve the issue of non-differentiability. As a result, you have to remove the L1 term from HingeGradient, and use the Breeze's OWLQN l

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user loachli commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-69555498 I have tested QWLQN in spark 1.1. Based on org.apache.spark.mllib.optimization.LBFGS, I create another class org.apache.spark.mllib.optimization.QWLQN. The main change i

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dbtsai commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68982905 The LBFGS solver will just optimize it no matter what; however, when LBFGS tried to approximate the hessian near zero, the whole problem is not well defined. You probably

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user loachli commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68982538 @dbtsai I used SVMWithLBFGS to test and found the function HingeGradient.compute would return 0 sometimes. But SVMWithLBFGS did not report any issues and I could still

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user dbtsai commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68912504 Since hinge loss in SVM is not differentiable around zero, L-BFGS will not work correctly. You need to use OWLQN to address the non-differentiable issue. PS, I tried to ad

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user AmplabJenkins commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68875919 Test PASSed. Refer to this link for build results (access rights to CI server needed): https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/25

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68875907 [Test build #25102 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25102/consoleFull) for PR 3890 at commit [`ddda43e`](https://gith

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68867190 [Test build #25102 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25102/consoleFull) for PR 3890 at commit [`ddda43e`](https://githu

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user AmplabJenkins commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68793049 Test PASSed. Refer to this link for build results (access rights to CI server needed): https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/25

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68793032 [Test build #25062 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25062/consoleFull) for PR 3890 at commit [`19852dc`](https://gith

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user SparkQA commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68783093 [Test build #25062 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/25062/consoleFull) for PR 3890 at commit [`19852dc`](https://githu

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user mengxr commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68782440 ok to test --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabl

[GitHub] spark pull request: [MLLib]SPARK-5027:add SVMWithLBFGS interface i...

Github user mengxr commented on the pull request: https://github.com/apache/spark/pull/3890#issuecomment-68782432 add to whitelist --- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature