Description:

Indefinite learning problems occur frequently if non-metric proximity measures are used (some neural network kernels, dynamic timewarping measures, alignment functions, inner distance and many other). The respective (supervised) learning algorithms have often quadratic to cubic complexity and a non-sparse decision function.

In this library a Krĕin space Core Vector Machine (iCVM) solver is derived. A sparse model with linear runtime complexity can be obtained under a low rank assumption. The obtained iCVM models can be applied to indefinite kernels without additional preprocessing. Using iCVM one can solve CVM with usually troublesome kernels having large negative eigenvalues or large numbers of negative eigenvalues.

In addition to the referenced paper the code provides an effective sparsification approach such that the final model is sparse again.

Changes to previous version:

Some tiny errors in the Nystroem demo scripts - should be ok now Initial Announcement on mloss.org.

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Linux

Data Formats: Hdf, Csv

Tags: Large Scale, Supervised Learning, Non Mercer, Indefinite Kernels, Core Vector Machine

Archive: download here

Comments

No one has posted any comments yet. Perhaps you'd like to be the first?

Leave a comment

You must be logged in to post comments.