ICML ML Evaluation 2009 - OpenResearch.org

Carlos Santos's personal annotations on this page

Carlos_santos bookmarked on 2009-04-02 icml MachineLearning evaluation research

this workshop intends to continue the debate within the
machine learning community into how we evaluate new algorithms.

this workshop intends to continue the debate within the
machine learning community into how we evaluate new algorithms.
In the course of three previous workshops, the debate has evolved to focus around
the following issues which have captured the interest of the community:

* the role of experiments in evaluation

* the use of one, community wide, evaluation measure (e.g., Accuracy, AUC,
F-measure)

* the relevance of statistical tests to evaluation

* the effectiveness of the UCI data sets for evaluation

* the need for sharing and characterizing benchmark data sets in general

* how to promote the views of this workshop to the rest of the community
The 2008 ICML workshop concluded with agreement that we, as a scientific community,
should substantially change how evaluation is performed in machine learning. We,
however, disagreed on the direction that this change should take.
we plan to have several invited speakers. Some, from
outside of our research community, will be able to criticize our accepted
practices from an external point of view. Some, from inside our community,
will discuss how we could improve on our current practices.

This link has been bookmarked by 1 people . It was first bookmarked on 02 Apr 2009, by Carlos Santos.