Machine learning library

[KetilMalde]

Note that this was proposed some years ago, and any prospective student should first identify a suitable mentor, and discuss the details with her.

Machine learning includes many methods (e.g, neural networks, support vector machines, hidden markov models, genetic algorithms, self-organizing maps) that are applicable to classification and/or pattern recognition problems in many fields.

A selection of these methods should be implemented as a Haskell library, the choice would depend on the qualifications and interests of the student. Optionally, a more limited library with an application using it could be implemented.

My main interest is in bioinformatics, but as machine learning methods are useful in a vast number of fields, I'm happy to hear from prospective mentors and other people who are interested, as well as from prospective applicants.

Previously Interested Students

• Anil Vaitla <avaitla16@…> (Matrix Decompositions, SVM, and HMM)
• Andreas Launila < http://lokoin.org/contact> (SVM and HMM)
• Jiri Hysek (dvekravy) <xhysek02@…> (NN and GA)
• Charles Blundell <blundellc@…> (SVM, HMM, ID3/C4.5; already done NN, Q-learn, SOM, ~GA)
• P McArthur < http://www.dysfunctor.org/about/> (Hidden Markov Model)
• Dave Tapley <dukedave@…> (Studying: < http://www.cse.dmu.ac.uk/msccir/structure.html>)
• Ivan Dilchovski <root.darkstar@…> (currently doing graduation paper on NN)
• Dinesh G<g.dinesh.cse@…> (SVM, PCA, SVD, Random Projection, Cluterings techniques) Currently doing my masters in Indian Institute Of Science. Area Of Interest: Machine learning
• Moises Osorio <wcoder.mx@…> (Genetic algorithms)

Currently Interested Students

• Martin Boyanov <mboyanov@…> (HMM, k-means clustering, Naive Bayes,neural networks)

[lokorin]

Added myself to the list of interested students.

[Moridin]

Added myself to the interested students too.

[g_dinesh_cse]

Replying to KetilMalde:

Machine learning includes many methods (e.g, neural networks, support vector machines, hidden markov models, genetic algorithms, self-organizing maps) that are applicable to classification and/or pattern recognition problems in many fields. A selection of these methods should be implemented as a Haskell library, the choice would depend on the qualifications and interests of the student. Optionally, a more limited library with an application using it could be implemented. My main interest is in bioinformatics, but as machine learning methods are useful in a vast number of fields, I'm happy to hear from prospective mentors and other people who are interested, as well as from prospective applicants. == Interested Mentors == == Interested Students == * Andreas Launila < http://lokoin.org/contact> (SVM and HMM) * Jiri Hysek (dvekravy) <xhysek02@…> (NN and GA) * Charles Blundell <blundellc@…> (SVM, HMM, ID3/C4.5; already done NN, Q-learn, SOM, ~GA) * P McArthur < http://www.dysfunctor.org/about/> (Hidden Markov Model) * Dave Tapley <dukedave@…> (Studying: < http://www.cse.dmu.ac.uk/msccir/structure.html>) * Ivan Dilchovski <root.darkstar@…> (currently doing graduation paper on NN)

[mitar]

I recommend also naive Bayes classifier.

For implementation in Python of many of those algorithms you could check  Orange.

[carette]

I have a student working on something like this already. What is the best way to coordinate?

[mboyanov]

Added myself to list of interested students.

I recently had to do some machine learning on matlab and I spent countless hours debugging due to the lack of type errors. From a programmer's point of view,implementing these algorithms in Haskell would allow users to have more control over their data.

Do you think that there is a need for this in Haskell,seeing that the project has been open for so long? What algorithms do you think should be implemented? Also, is there a chance that this project be combined with  /A statistics library and environment?

[mcandre]

I wrote a simple genetic algorithms API, would that help?

 https://github.com/mcandre/genetics