mloss.org libAGFhttp://mloss.orgUpdates and additions to libAGFenSat, 23 Jan 2016 23:46:19 -0000libAGF 0.9.8http://mloss.org/software/view/257/<html><p>This software is written in C++ and contains routines for statistical classification, probability estimation and interpolation/non-linear regression. Two variable bandwidth kernel methods are adopted: k-nearest neighbour (KNN), and a balloon estimator based on Gaussian kernels, hence Adaptive Gaussion Filtering (AGF). A library of easy-to-use, single-call functions (you call a single function once for each estimate--no initialization required) are included, as well as command-line executables.
</p>
<p>The statistical classification routines are particularly powerful, allowing you to generate a pre-trained model by searching for the class borders. These can then be used to make rapid classifications which nonetheless return estimates of the conditional probabilities.
</p>
<p>Clustering routines are a recent addition.
</p></html>Peter MillsSat, 06 Dec 2014 02:35:39 -0000http://mloss.org/software/rss/comments/257http://mloss.org/software/view/257/clusteringnonparametric density estimationsupervised learninginterpolationinverse methodskernel estimationnonlinear regressionprobability estimationstatistical classification<b>Comment by Peter Mills on 2012-03-15 05:04</b>http://mloss.org/comments/cr/14/257/#c672<p>I had hoped to have multi-class border-classification ready by now, but the simple generalization I had envisioned to implement it won't work in all cases. The idea was to use matrix inversion to solve for the conditional probabilities, but quite obviously (in retrospect) you can solve for the class without being able to determine all the conditional probabilities. Likely we need two cases: one where all the conditional probabilities can be found, and one where only that of the retrieved class can be found and these two cases need to interoperate. A recursive or hierarchical model would seem to be the best solution here.<br />
</p>
<p>I realize that there is literature relating to the problem of creating multi-class classifications from two-class, however I do not currently have access to commercial journals as I am not affiliated with an academic or research institution. It is also an enjoyable challenge to try and figure these things out for yourself, from scratch, so to speak.
</p>
<p>Likewise I had hoped to have the optimal-bandwidth Gaussian PDF estimation ready. I had made some progress on it, but the test cases were not giving consistent results and I have failed to work on it in the intervening months.
</p>Peter MillsThu, 15 Mar 2012 05:04:16 -0000http://mloss.org/comments/cr/14/257/#c672<b>Comment by Peter Mills on 2014-04-15 04:55</b>http://mloss.org/comments/cr/14/257/#c709<p>Multi-borders classification is now ready. I am very pleased (and pleasantly surprised) with how well it works.
</p>Peter MillsTue, 15 Apr 2014 04:55:05 -0000http://mloss.org/comments/cr/14/257/#c709<b>Comment by Peter Mills on 2016-01-23 23:46</b>http://mloss.org/comments/cr/14/257/#c762<p>The libAGF library has been combined with two other libraries and moved to Github under the project, libmsci:
https://github.com/peteysoft/libmsci
</p>Peter MillsSat, 23 Jan 2016 23:46:19 -0000http://mloss.org/comments/cr/14/257/#c762