Data, Knowledge & Life

Weiwei Cheng's blog

Archive for 五月 2011

A repository of datasets for monotone learning

leave a comment »

We are announcing a monotone learning data repository. The datasets can be downloaded at our KEBI website:

http://www.uni-marburg.de/fb12/kebi/research/repository/monodata

Monotone learning is an important research task in machine learning community and has a wide range of applications. Often, monotone learners lead to more satisfactory predictive performance; and sometimes monotonicity is even required for applications in medical diagnosis, security system design, etc. For example, a system to predict the patient’s overall wellness should have an outcome that is monotonic with respect to the toxicity measure of that patient. No medical doctor will accept a learning model violating this monotonicity.

In our group, we are working on some new ideas for monotone learning. Surprisingly, during our ongoing research we’ve found, despite a tremendous interest on this topic, there is still no public available dataset collection for monotone learning. We’ve worked hard on collecting some benchmark datasets, and we believe it is beneficial to all researchers interested in this topic that we make our collection publicly available.

This datasets collection is still preliminary. There are a number of points on our todo list. In case you have any comment, or if yourself have a monotone learning dataset that you would like to add, please let us know.

Written by Weiwei

05/05/2011 at 00:23

发表在 学术