OpenCV Q&A Forum - RSS feedhttp://answers.opencv.org/questions/OpenCV answersenCopyright <a href="http://www.opencv.org">OpenCV foundation</a>, 2012-2018.Wed, 12 Aug 2020 07:40:52 -0500Why vector module doesn't grow up with not regularized logistic regressionhttp://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/ I'm training on a linearly separable dataset, without regularization, with thousands of iterations and a leraning rate of 0.001
(but in fact I have used several combinations, and with several datasets)
Taking into account that the dataset is clearly linearly separable I would expect a result as:
100% of accuracy
an increasing module of the vector defined by the learnt weights (discarding first one)
But, comparing the learnt weights when using L2 regularization or not, the changes in learn weights only happens in the fourth or fifth digit...
Is that correct?
Wed, 12 Aug 2020 05:53:33 -0500http://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/Comment by berak for <p>I'm training on a linearly separable dataset, without regularization, with thousands of iterations and a leraning rate of 0.001
(but in fact I have used several combinations, and with several datasets)</p>
<p>Taking into account that the dataset is clearly linearly separable I would expect a result as:
100% of accuracy
an increasing module of the vector defined by the learnt weights (discarding first one)</p>
<p>But, comparing the learnt weights when using L2 regularization or not, the changes in learn weights only happens in the fourth or fifth digit...</p>
<p>Is that correct?</p>
http://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/?comment=233571#post-id-233571also: https://answers.opencv.org/question/233267/termcriteria-in-logisticregression/Wed, 12 Aug 2020 07:40:52 -0500http://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/?comment=233571#post-id-233571Comment by berak for <p>I'm training on a linearly separable dataset, without regularization, with thousands of iterations and a leraning rate of 0.001
(but in fact I have used several combinations, and with several datasets)</p>
<p>Taking into account that the dataset is clearly linearly separable I would expect a result as:
100% of accuracy
an increasing module of the vector defined by the learnt weights (discarding first one)</p>
<p>But, comparing the learnt weights when using L2 regularization or not, the changes in learn weights only happens in the fourth or fifth digit...</p>
<p>Is that correct?</p>
http://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/?comment=233570#post-id-233570> an increasing module of the vector
can you explain "module" here ?
can you show your code ? also, a mini dataset for reproducing your problem will be usefulWed, 12 Aug 2020 07:30:16 -0500http://answers.opencv.org/question/233569/why-vector-module-doesnt-grow-up-with-not-regularized-logistic-regression/?comment=233570#post-id-233570