Data Mining

PyMF: Python Matrix Factorization Module

pymf.png

I put together some code I wrote earlier this year into a matrix factorization library for Python http://pymf.googlecode.com. PyMF currently includes methods for Non-negative matrix factorization (NMF), Convex non-negative matrix factorization (CNMF), Semi non-negative matrix factorization (SNMF), Archetypal analysis (AA), Simplex volume maximization (SiVM), Convex-hull non-negative matrix factorization (CHNMF), Binary matrix factorization (BNMF), Singular value decomposition (SVD), Principal component analysis (PCA), K-means clustering (Kmeans), CUR decomposition (CUR), and Compaxt matrix decomposition (CMD). Descriptions of the methods and links to the papers are included in the code. The library is not very well tested and probably needs some beautification ;). Most methods should work fine with data stored in hdf5 tables (see h5py for details).

CIKM 2010 paper accepted

Our paper Yes We Can - Simplex Volume Maximization for Descriptive Web-Scale Matrix Factorization got accepted at CIKM 2010 as a short paper. Out of the 945 submissions, 127 (13.4%) were accepted as full papers and 169 (17.9%) as short papers.

CIG 2010 paper accepted + tutorial + special session

Membership statistics on various MMORPG, according to mmogchart.com.

Our CIG 2010 (IEEE Conference on Computational Intelligence and Games) paper on Analyzing
the Evolution of Social Groups in World of Warcraft
got accepted, acceptance rate is 50%.

At CIG, I will give a tutorial on Game Mining – Data Mining in Games at CIG, and together with people from KD and VSM we (Chistian Bauckhage, Olana Missura, Thomas Gaertner, and Kristian Kersting) organize a special session on Game Mining.

Paper selected as one of the best at ICDM 2009

Our paper on convex non-negative matrix factorization in the wild has been selected as on of the best papers at the Int. Conf. on Data Mining 2009 (we won a possible publication in Knowledge and Information Systems (KAIS), nice :-)

ICDM 2009 paper accepted

Convex Non-Negative Matrix Factorization applied to guild information in the MMORPG World of Warcraft

Our ICDM 2009 paper on Convex Non-Negative Matrix Factorization in the Wild got accepted as a regular paper. Acceptance rate for regular papers is 8.9%. From 786 submissions 140 were selected for presentation, 70 as a regular paper and the remaining 70 as short papers.

ICSC 2009 paper accepted

Our ICSC 2009 paper on Archetypal Images in Large Photo Collections got accepted. Acceptance rate is 30%.

DAGM 2009 paper accepted

DAGM paper on archetypal analysis accepted.


archetypes