Installation:

ALsvm and AWsvm are "Spider" objects for matlab, you need
to download the Spider main package from its website
http://www.kyb.tuebingen.mpg.de/bs/people/spider/
and install it. 

Now download 
http://www.kyb.mpg.de/bs/people/pgehler/mil/milsvm.zip
and unzip the files. If you are also interested in the benchmark 
datasets you can obtain them from 
http://www.kyb.mpg.de/bs/people/pgehler/mil/milbenchmark.zip
and unzip them into the same directory. This will create
a new directory e.g. 'MILsvm/MILdata' and a routine
'load_mildata.m' to load the datasets into matlab.
(The original datasets are obtained from 
http://www.cs.columbia.edu/~andrews/mil/datasets.html
I host them here only for easier use with my code.)

Copy the file in 'MILsvm/spider/log.m' to the following
subdirectory of your Spider installation:
'SPIDERROOT/basic/@algorithm/'

The files 'sampleRunToydata.m' and 'sampleRunBenchmark.m' 
provide a step by step guidance of how to use the software. 
If you have not used spider before it is recommended you
start to have a look at 'sampleRunToydata.m' first. This 
routine work on the same toy data set as presented in the 
"Deterministic Annealing for MIL" paper.



If you have a question regarding the package or its installation
please do not hesitate to write me an email, I would be 
happy to hear if the software works for you so I can 
improve it and the description. 

Peter Gehler - peter.gehler@tuebingen.mpg.de