Introduction
SVMheavy is yet another SVM implementation, intended for use as a testbed for comparing different incremental training algorithms. It is based around a generic optimisation core which is then used to implement both pattern recognition and regression. At present it supports active set training (as per "Incremental Training of Support Vector Machines", A. Shilton, M. Palaniswami, D. Ralph, A. C. Tsoi, submitted to IEEE Transactions on Neural Networks), Platt's SMO algorithm and simple gradient descent.
The code was written by Alistair Shilton, University of Melbourne, Electrical and Electronic Engineering department (apsh@ecr.mu.oz.au). Currently there is no real documentation, but we do plan to write some eventually (compilation instructions can be found in the readme file in the zip). The code has been tested in windows (compilation using DJGPP) and unix (g++) and has no killer bugs that we are aware of, but use at own risk.
SVMheavy may be downloaded either in binary form (windows only) or as source
code. Enjoy.
Download SVMheavy sourcecode.
Download SVMheavy pattern recognition binary (windows).
Download SVMheavy regression binary (windows).
Download SVMheavy pattern recognition binary with flop counting (windows).
Download SVMheavy regression binary with flop counting (windows).