Re: [Scikit-learn-general] ELM and Deep Learning

There are other more specialised projects that facilitate modular neural networks. The idea in scikit-learn is to provide useful out-of-the-box components for well-established solutions to certain types of tasks that fit a simple interface. This often means limiting their flexible use from the pers

[Scikit-learn-general] ELM and Deep Learning

Hi, I have gone through documentation of Pull 3306. I am glad that ELM will soon be part of scikit-learn. But It is just working as an simple ML algorithm, which can be fitted to data and can predict based on the trained model. I was considering to develop something different. My plan is to implem

Re: [Scikit-learn-general] [GSoC 2015] Cross-validation and Meta-Estimators for semi-supervised learning

Hi Boyuan, It's good to hear you're an experienced scikit-learn user, and that it has worked for you. It's also pleasing to hear someone's interested in this project, because I feel the semi-supervised capabilities of scikit-learn and its API have been left half-baked. I strongly recommend, howev

Re: [Scikit-learn-general] [GSoC] Metric Learning

I don't know a lot about metric learning either, but it sounded like from your initial statement that fit(X, D) where D is the target/known distance between each point in X might be appropriate. I have no idea if this is how it is formulated in the literature (your mention of asymmetric metrics mea

[Scikit-learn-general] [GSoC 2015] Cross-validation and Meta-Estimators for semi-supervised learning

Hi all: I am a Master's student in the European Union's Erasmus Mundus LCT program, studying natural language processing at Saarland University, Germany and also doing machine learning and information retrieval at Max-Planck Institute for Informatics, which is on the same campus. These years

Re: [Scikit-learn-general] [GSoC] Metric Learning

In summary, I think this does look like a good basis for a proposal :) On 03/18/2015 05:14 PM, Artem wrote: ​ Do you think this interface would be useful enough? ​One of mentioned methods (LMNN) actually uses prior knowledge in exactly the same way, by comparing labels' equality. Tho

Re: [Scikit-learn-general] [GSoC] Metric Learning

> > ​ > Do you think this interface would be useful enough? ​One of mentioned methods (LMNN) actually uses prior knowledge in exactly the same way, by comparing labels' equality. Though, it was designed to facilitate KNN. ​ ​ ​Authors of the other one (ITML) explicitly mention in the paper that on

Re: [Scikit-learn-general] [GSoC] Metric Learning

On 03/18/2015 02:53 PM, Artem wrote: I mean that if we were solving classification, we would have y that tells us which class each example belongs to. So if we pass this classification's ground truth vector y to metric learning's fit, we can form S and D inside by saying that observations from

Re: [Scikit-learn-general] [GSoC] Metric Learning

I mean that if we were solving classification, we would have y that tells us which class each example belongs to. So if we pass this classification's ground truth vector y to metric learning's fit, we can form S and D inside by saying that observations from the same class should be similar. ​ > On

Re: [Scikit-learn-general] [GSoC] Metric Learning

The issue with having anything else than fit(X, y) would break cross_val_score, GridSearchCV and Pipeline. I agree that more control is good, but having functions that don't work well with the rest of the package is not great. Only being able to "transform" to a distance to the training set is

[Scikit-learn-general] PhD student position on Machine Learning for NLP in Nantes (France)

Hi, Jurismarchés is a Nantes (France) based company selecting call for tenders for its clients (business opportunity monitoring). Jurismarchés propose a PhD in a mix location : university lab (IRISA in Vannes) and company (a french convention known as "Thèse CIFRE"). The subject would be on using

Re: [Scikit-learn-general] [GSoC] Metric Learning

Well, we could go with fit(X, y), but since algorithms use S and D, it'd better to give user a way to specify them directly if (s)he wants to. Either way, LMNN works with raw labels, so it doesn't require any changes to the existing API. On Wed, Mar 18, 2015 at 7:26 PM, Gael Varoquaux < gael.varoq

Re: [Scikit-learn-general] [GSoC] Metric Learning

On Wed, Mar 18, 2015 at 07:21:18PM +0300, Artem wrote: > As to what y should look like, it depends on what we'd like the algorithm to > do. We can go with usual y vector consisting of feature labels. Actually, LMNN > is done this way, the optimization objective depends on the equality of labels > o

Re: [Scikit-learn-general] [GSoC] Metric Learning

Yeah, the API is the most important question of the implementation. These learners are not classifiers (though there exist metric-adapting algorithms like Neighbourhood Components Analysis ), so they don't fit into usual estimator-lik

Re: [Scikit-learn-general] [GSoC] Metric Learning

On Wed, Mar 18, 2015 at 11:32:05AM -0400, Andreas Mueller wrote: > Do you have an idea of what y would look like? Me. Not sure, no. I haven't looked at the corresponding literature. G > Also +1 on what you said (but you knew that ;) > On 03/18/2015 11:27 AM, Gael Varoquaux wrote: > > Simple, ef

Re: [Scikit-learn-general] [GSoC] Metric Learning

Do you have an idea of what y would look like? Also +1 on what you said (but you knew that ;) On 03/18/2015 11:27 AM, Gael Varoquaux wrote: > Simple, efficient and robust metric learning that learns on a supervised > set and can do a transform that applies the metric? Do you think that > would be

Re: [Scikit-learn-general] [GSoC] Metric Learning

Simple, efficient and robust metric learning that learns on a supervised set and can do a transform that applies the metric? Do you think that would be useful? It seems to me that it would. If people agree that it would be useful with such a very simple API, I would be in favor of a GSoC proposal

Re: [Scikit-learn-general] [GSoC] Metric Learning

Hey. I am not very familiar with the literature on metric learning, but I think one thing that we need to think about before is what the interface would be. We really want something that works in a .fit().predict() or .fit().transform() way. I guess you could do "transform" to get the distances

Re: [Scikit-learn-general] Maximum Likelihood Classifier

Hi Pietro. Can you explain what you mean by maximum likelihood classifier? I think it may refer to classifiers based on generative models (we have several naive Bayes classifiers and linear and quadratic discriminant analysis). Cheers, Andy On 03/18/2015 04:03 AM, Pietro wrote: > Dear all, > > I

[Scikit-learn-general] [GSoC] Metric Learning

Hello everyone Recently I mentioned metric learning as one of possible projects for this years' GSoC, and would like to hear your comments. Metric learning, as follows from the name, is about learning distance functions. Usually the metric that is learned is a Mahalanobis metric, thus the problem

[Scikit-learn-general] ELM and Deep Learning

Thanks Joel for clarification. I have gone through documentation of Pull 3306. I am glad that ELM will soon be part of scikit-learn. But It is just working as an ML algorithm, which can be fitted to data and can predict based on the trained model. I was considering to develop something different.

[Scikit-learn-general] Scikit-learn training in 2 weeks in London (commercial training) + PyDataLondon Kaggle meet

Hello. On April 7-8th in London I'm helping to run a 2 day scikit-learn/statsmodels/pandas commercial training course. It is led by an ex-Googler, it is aimed at engineers and less experienced data scientists who want to firm up their statistical background and machine learning abilities. It'll be

[Scikit-learn-general] Maximum Likelihood Classifier

Dear all, I'm looking for a Maximum Likelihood Classifier. I've looked in the documentation but I didn't found it, Is the Maximum Likelihood Classifier not implemented in scikit-learn, or I've missed something? If it is missing do you have any library to suggest? I found it in Mlpy, but the last