12.01
CLOP is a software package containing several ready-to-use machine learning algorithms which is developed during Performance Prediction Challenge, as a part of Model Selection Workshop in IEEE World Congress on Computational Intelligence, Vancouver, BC, Canada, July 2006. For more information and for related citations, please refer to the following paper:
Isabelle Guyon, Amir Saffari, Gideon Dror, and Joachim Buhmann, Performance Prediction Challenge, In Proceedings of International Conference on Neural Networks (IJCNN), IEEE World Congress on Computational Intelligence (WCCI) 2006, Pages: 2958-2965, Vancouver, British Columbia, Canada, July 2006.
CLOP is based on SPIDER package from Department of Empirical Inference for Machine Learning and Perception, Max Planck Institute for Biological Cybernetics, Tuebingen, Germany. CLOP has more algorithms provided by challenge organizers compared to Spider and it runs on MATLAB, but it does not depend on any particular toolbox of MATLAB. Please feel free to contact us if you had any problem, suggestion, need more information.
Download the latest version from here:
Version 1.0: (Release date: May 17, 2006)
Powered by FireStats
[...] CLOP Version 1.0 is ready to use. [...]
Hi Amir,
Your toolbox CLOP is really a great works. Thanks very much for this excellent package.
Hi,
I tried to use your package however I wonder what is the data file format. I could not find in any documents that how the files and file formats should be. I am actually try to use your package for feature selection. I almost check all feature dataset fie format but non of them works. I got the following error and so far I couldn\’t find what is the reason.
??? Error using ==> fscanf
Invalid file identifier. Use fopen to generate a valid file identifier.
Error in ==> read_parameters at 12
p.data_type=fscanf(fp, \’Data type: %s\\n\’);
Error in ==> create_data_struct at 19
p = read_parameters(filename);
Hi,
In fact, you can use any sort of files as you like, just load your data into MATLAB and create a data object. All methods work with data objects. However, if you want to use the leading functions which comes in sample code section, try to download one of the challenge files and see it’s format, e.g. http://www.modelselect.inf.ethz.ch/datasets.php
Cheers
Amir
Hi Amir,
is it possible to use this package for supervised multi-class classification problems?
Have you ever tried?
Hi Diego,
As far as I know, it’s been designed for binary, but you can easily write a 1-vs-all multi-class classifier for it.
Cheers, Amir