Re: [Scikit-learn-general] GSOC

LSTMs have proven to be useful as sequence learning methods, in some cases outperforming HMMs[1]. But I know that there was some discussion about removing the HMMs from scikit-learn, so I presume that this and the fact that they require a more DSL-like configuration would not make them a good GSOC

Re: [Scikit-learn-general] Distributed RandomForests

Thank you Peter, I found that the feature extraction was taking a lot of extra memory and that was not related to wiseRF, so you were right. Actually, from "top" it seems the training part was taking only an extra 20% of memory than the size of the dataset itself, wich is pretty impressive. So at t

Re: [Scikit-learn-general] ICML'13 challenge : blackbox learning

2013/4/26 Eustache DIEMERT : > I tried a very naive bootstrapping approach (learn RF on supervised data, > predict on unsupervised, then learn RF on all) but with no luck either. Sounds like you've now got a self-training algorithm with only one iteration. You may have more luck with a proper self

Re: [Scikit-learn-general] LibSVM GUI error

On Fri, Apr 26, 2013 at 05:20:48PM +0530, Shishir Pandey wrote: > This error is for sklearn version 0.13.1 I am assuming six.move might be > added to 0.14 version. Because the link I used was for the dev version > of sklearn - > http://scikit-learn.org/dev/auto_examples/applications/svm_gui.html

[Scikit-learn-general] Machine Learning Cheat sheet flow chart for docs

Hi everyone. I recently submitted a pull-request to add an interactive version of Andy's machine learning flowchart (see here [1] for those who missed it) to the documentation. I'd very much like to hear what

Re: [Scikit-learn-general] LibSVM GUI error

I created a issueticket for this at https://github.com/scikit-learn/scikit-learn/issues/1901 2013/4/26 Shishir Pandey > I get the following error: > > Traceback (most recent call last): >File "C:\Users\xyz\ml\svm_gui.py", line 30, in > from sklearn.externals.six.moves import xrange >

[Scikit-learn-general] ICML'13 challenge : blackbox learning

Hi fellow sklearners, I'm casually participating in the ICML'13 challenge on blackbox learning [1] using sklearn and wanted to report progress and seek new ideas on how sklearn tools can be used for this task. Some basic info about the dataset: - we don't know the meaning of features nor predicti

Re: [Scikit-learn-general] LibSVM GUI error

2013/4/26 Shishir Pandey : > This error is for sklearn version 0.13.1 I am assuming six.move might be > added to 0.14 version. Because the link I used was for the dev version > of sklearn - > http://scikit-learn.org/dev/auto_examples/applications/svm_gui.html#example-applications-svm-gui-py You sh

Re: [Scikit-learn-general] LibSVM GUI error

I get the following error: Traceback (most recent call last): File "C:\Users\xyz\ml\svm_gui.py", line 30, in from sklearn.externals.six.moves import xrange ImportError: No module named six.moves This error is for sklearn version 0.13.1 I am assuming six.move might be added to 0.14 versi

Re: [Scikit-learn-general] MiniBatchKMeans doesn't really re-run its algorithm 'n_init' times

On 04/26/2013 12:45 PM, Jaques Grobler wrote: > Issue opened https://github.com/scikit-learn/scikit-learn/issues/1900 > Thank you. I might attempt at writing a documentation patch myself, if nobody beats me at it (no promise though). Best regards, Stefano --

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

>From what you are saying, the independent variables are the parameters of the cost function. It is your search space, right? If you change the scale, of course the gradient descent behavior will be different. Also, if the input parameters are scaled properly, (let's say that the variables that had

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

On Fri, Apr 26, 2013 at 04:17:36PM +0530, Shishir Pandey wrote: > @Jaques Grobler: I ran the libsvm GUI code on the sklearn version 13.1 > it was giving error importing - "from sklearn.externals.six.move import > xrange". Which error? Could you copy/paste it here? G -

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

@Jaques Grobler: I ran the libsvm GUI code on the sklearn version 13.1 it was giving error importing - "from sklearn.externals.six.move import xrange". But I commented the above line and it is working just fine. As you have suggested GUI example might not really be that necessary. Illustrating dif

Re: [Scikit-learn-general] MiniBatchKMeans doesn't really re-run its algorithm 'n_init' times

Issue opened https://github.com/scikit-learn/scikit-learn/issues/1900 2013/4/26 Jaques Grobler > I'll open up an issue > > > 2013/4/26 Gael Varoquaux > >> > This different behaviour is not made apparent in the documentation, >> > either. >> >> > So, my question is: is this a bug, or is it inte

Re: [Scikit-learn-general] MiniBatchKMeans doesn't really re-run its algorithm 'n_init' times

I'll open up an issue 2013/4/26 Gael Varoquaux > > This different behaviour is not made apparent in the documentation, > > either. > > > So, my question is: is this a bug, or is it intended behaviour? > > I believe that it is an intended behaviour: the MiniBatch is meant to do > only a few pass

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

@Shishir Pandey on a slight tangent, what problems are you having with running Libsvm GUI? I wonder if a GUI interactive example would really be necessary - we could just have an example illustrating the difference with plots when data is not scaled or scaled.. if people find that useful. But the

Re: [Scikit-learn-general] MiniBatchKMeans doesn't really re-run its algorithm 'n_init' times

> This different behaviour is not made apparent in the documentation, > either. > So, my question is: is this a bug, or is it intended behaviour? I believe that it is an intended behaviour: the MiniBatch is meant to do only a few passes on the data, for efficiency issues. However, the documenta

[Scikit-learn-general] MiniBatchKMeans doesn't really re-run its algorithm 'n_init' times

Hello scikit-learn developers. I've noticed a somewhat unexpected difference between the behaviour of the KMeans class and the MiniBatchKMeans class. When the 'n_init' argument is given, I'd expect both of these classes to run the corresponding algorithm (Lloyd and mini-batch k-means, respectivel

Re: [Scikit-learn-general] Participation in GSoC 2013

Hi, Here's an update about my possible participation in GSoC this year. If they have time, I think that Gilles Louppe or Arnaud Joly would make good mentors since they are from University of Liège too. Gilles wrote his master thesis on recommender systems so he has good knowledge of matrix fact

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

(first-order) GD uses a single learning rate for all features - if features have a different variability its difficult to find a one-size-fits-all learning rate - the parameters of high variability features will tend to oscillate whereas the parameters of low variability features will converge too

Re: [Scikit-learn-general] Effects of shifting and scaling on Gradient Descent

afaik fits tend to work better and so do classifiers. it's much easier to have a classifier try to fit between -1 and 1 then 0 and 1 so it also helps convergence. http://stats.stackexchange.com/questions/41704/how-and-why-do-normalization-and-feature-scaling-work and then http://en.wikipedia.