This is a library of classifiers put together by Andy Port for his own self-enrichment.
All datasets are formatted in a similar way, as a MATLAB .mat file containing a tensor X containing sample points (indexed by the first dimension) and a column vector y containing labels.
With the exception of small datasets (e.g. the USPS dataset of handwritten digits), you must use the shell scripts get_*.sh to download (and format) each dataset. Note: If you're on Windows, check out the README.txt files in each dataset's directory or just go through the steps in the get_*.sh files -- all these scripts do is: download the compressed datasets, decompress/decode them, then run some python and/or octave scripts to reformat them to the above dataset formatting standard.
Navigate to Classifiers2LearnWith/experiments directory, then run whichever you like.
When coding this I used Octave 4.2.0-rc2 ... I don't think you'll have any problems compiling the contained .m files with MATLAB, except that I have some suspicion that matlab does not have a built-in argv() function to grab CLI inputs.