Dear Pascal Researchers,
I’m glad to announce that uploading data sets to mldata will count
towards your Activity score within Pascal.
mldata.org is a web site which has been funded by Pascal2 and which
allows you to upload and share your data sets. As data sets are
central to machine learning, we hope to further machine learning
research through this web site. You may upload data sets in any format
you like, but there also exists special support for uploading data
sets in one of the formats that mldata can parse (in particular, ARFF)
To accomodate such advanced features, mldata is a bit more structured
than a mere repository of your data files, but you shouldn’t be afraid
of that. Basically, there are four types of entries:
– Data (raw data)
– Task (learning tasks, for now mainly supervised learning)
– Method (a ML method to be applied to tasks)
– Challenge (a collection of learning tasks to form a learning challenge)
mldata also has support for collaborative editing, such that others
can improve your description, a forum for discussing topics, and so on.
So if you have some data set from one of your papers and would like to
share it with other researchers, make sure to have a look at mldata.
The number of uploaded data sets will then count towards your
Activitiy score which determines your base site funding for the last
project period. How and when we will determine this exactly will be
announced in a later email. We will probably have a look at your data
around the end of March 2012.
One of the first question usually ask is about licensing. We have put
a number of Open Source licenses plus discussion here: