Re: [scikit-learn] Markov Clustering?

Thanks for pointing that out, I sort of picked it up by word of mouth so I'd assumed it had a bit more precedence in the academic world. I'll look into it a little more, but I'd definitely be interested in contributing something else if that doesn't work out. -Allan On Sat, Dec 3, 2016 at 4:45 P

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

As long as the feature ordering has a meaningful spatial component (as is almost always the case when you are dealing with raw pixels as features) CNNs will almost always be better. CNNs actually have a lot fewer parameters than MLPs (depending on architecture of course) because of weight sharing

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

No, I am not saying it is better than CNN, but my images aren't real-life images but computer generated silhouettes. So CNN seemed to be overkill. I'll revisit CNN. I resized the images and converted it to grayscale. Now I am feeding [1,4800] now and I am getting good output with MLP. I looped over

Re: [scikit-learn] Markov Clustering?

Hey Allan. None of the references apart from the last one seems to be published in a peer-reviewed place, is that right? And "A stochastic uncoupling process for graphs" has 13 citations since 2000. Unless there is a more prominent publication or evidence of heavy use, I think it's disqualifie

Re: [scikit-learn] Markov Clustering?

Hey Andy, This algorithm does operate on sparse graphs so it may be beyond the scope of sci-kit learn, let me know what you think. The website is here , it includes a brief description of how the algorithm operates under Documentation -> Overview1 and Overview2. The referen

Re: [scikit-learn] Adding samplers for intersection/Jensen-Shannon kernels

Hi Valery. I didn't include them because the Chi2 worked better for my task ;) In hindsight, I'm not sure if these kernels are not to a bit too specialized for scikit-learn. But given that we have the (slightly more obscure) SkewedChi2 and AdditiveChi2, I think the intersection one would be a go

[scikit-learn] Adding samplers for intersection/Jensen-Shannon kernels

Hello, In the course of my work, I've made samplers for intersection/Jensen-Shannon kernels, just by small modifications to sklearn.kernel_approximation.AdditiveChi2Sampler code. Intersection kernel proved to be the best one for my task (clustering Docstrum feature vectors), so perhaps it'd b

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

On 12/03/2016 03:10 PM, Alekh Karkada Ashok wrote: Hey All, I chose MLP because they were images and I have heard MLPs perform better. Better than a convolutional neural net? Whoever told you that was wrong. I usually don't make absolute statements like this, but this is something that is p

Re: [scikit-learn] Markov Clustering?

Hi Allan. Can you provide the original paper? It this something usually used on sparse graphs? We do have algorithms that operate on data-induced graphs, like SpectralClustering, but we don't really implement general graph algorithms (there's no PageRank or community detection). Andy On 12/0

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

Hey All, I chose MLP because they were images and I have heard MLPs perform better. My application is detecting body parts from these images and therefore, the mapping would be pretty non-linear and this was my idea behind selecting MLP. Otherwise, I would have to engineer high dimension features

Re: [scikit-learn] Markov Clustering?

I don't think anyone is working on this. Contributions are always very welcome, but be aware before you start that the process of getting a completely new algorithm into scikit-learn will take a lot of time and reviews. On Sat, Dec 3, 2016 at 9:19 AM, Allan Visochek wrote: > Hi there, > > My nam

Re: [scikit-learn] Github project management tools

On 12/03/2016 01:20 PM, Nelle Varoquaux wrote: On 3 December 2016 at 10:08, Andy wrote: On 12/03/2016 12:26 PM, Raghav R V wrote: We could start with assigning priority labels like they use in numpy... That + milestones could help us prioritize? I feel milestones are too coarse. Or I'm us

Re: [scikit-learn] Github project management tools

On 3 December 2016 at 10:08, Andy wrote: > > > On 12/03/2016 12:26 PM, Raghav R V wrote: >> >> We could start with assigning priority labels like they use in numpy... >> That + milestones could help us prioritize? >> > I feel milestones are too coarse. Or I'm using them wrong. > And priority label

Re: [scikit-learn] Trying to get learning curves with custom scorer and leave one group out

That indeed looks odd. Can you reproduce with synthetic data? On 12/02/2016 10:40 PM, Matteo Niccoli wrote: My apologies, there was a typo in the code below, second example, should read: train_scores1, test_scores1 = validation_curve(SVC_classifier_LOWO_VC1, X, y, "C", parm_range1, cv =logo.sp

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

On 12/03/2016 05:29 AM, Gael Varoquaux wrote: On Sat, Dec 03, 2016 at 03:08:00PM +0530, Alekh Karkada Ashok wrote: I want use the Scikit-learn's MLPRegressor to map image to image. That is I have a numpy array of size [1000,2030400] (1000 samples, 76800x3 (RGB) pixels). Corresponding labelled

Re: [scikit-learn] Github project management tools

On 12/03/2016 12:26 PM, Raghav R V wrote: We could start with assigning priority labels like they use in numpy... That + milestones could help us prioritize? I feel milestones are too coarse. Or I'm using them wrong. And priority labels only work if people don't use the "high priority" all

[scikit-learn] Markov Clustering?

Hi there, My name is Allan Visochek, I'm a data scientist and web developer and I love scikit-learn so first of all, thanks so much for the work that you do. I'm reaching out because I've found the markov clustering algorithm to be quite useful for me in some of my work and noticed that there is

Re: [scikit-learn] Github project management tools

We could start with assigning priority labels like they use in numpy... That + milestones could help us prioritize? On Sat, Dec 3, 2016 at 11:52 AM, Gael Varoquaux < gael.varoqu...@normalesup.org> wrote: > On Fri, Dec 02, 2016 at 07:52:09PM -0500, Andy wrote: > > So did we ever decide on how to p

Re: [scikit-learn] Github project management tools

On Fri, Dec 02, 2016 at 07:52:09PM -0500, Andy wrote: > So did we ever decide on how to prioritize reviews? I don't know how to do this. > I think it might be helpful if Joel and me prioritize issues. I think that it would be useful. Although of course different people will have different priori

Re: [scikit-learn] Fwd: Scikit-learn MLPRegressor Help

On Sat, Dec 03, 2016 at 03:08:00PM +0530, Alekh Karkada Ashok wrote: > I want use the Scikit-learn's MLPRegressor to map image to image. That is I > have a numpy array of size [1000,2030400] (1000 samples, 76800x3 (RGB) > pixels). > Corresponding labelled images I have. Therefore Y is also [1000,2

[scikit-learn] Fwd: Scikit-learn MLPRegressor Help

Hi all, I want use the Scikit-learn's MLPRegressor to map image to image. That is I have a numpy array of size [1000,2030400] (1000 samples, 76800x3 (RGB) pixels). Corresponding labelled images I have. Therefore Y is also [1000,230400]. But according to documentation: *fit(X, y)* Fit the model to