Connectionists: Entropic affinities software available

[Max Vladymyrov]

Dear all,

We are pleased to advertise Matlab code for computing entropic affinities
as described in the following paper:

 M. Vladymyrov and M. A. Carreira-Perpinan: "Entropic affinities:
 properties and efficient numerical computation", ICML 2013.
 https://eng.ucmerced.edu/people/vladymyrov
 http://faculty.ucmerced.edu/mcarreira-perpinan/papers.html

The code efficiently and accurately computes Gaussian affinities for every
point in the dataset with a variable bandwidth sigma that corresponds to a
desired perplexity (effective number of neighbors) K.

The entropic affinities were originally introduced in:
 G. Hinton and S. T. Roweis: "Stochastic Neighbor Embedding", NIPS 2002.
Although these affinities have been mostly used with nonlinear embedding
algorithms, we believe they can give good results in many other problems
were Gaussian affinities are used, such as manifold learning, clustering
or semi-supervised learning. The code is available in the authors'
websites above.

Sincerely,
Max Vladymyrov and Miguel A. Carreira-Perpinan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20130717/118c1d9e/attachment.html>