Proceedings of the Twentieth International Conference on Machine Learning
This proceedings contains a written record of the Twentieth International Conference on Machine Learning (ICML-2003) held in Washington D.C. on August 21-24th, 2003. The response to the call for papers for this conference was strong. Out of 371 full-length research papers submitted, 96 were accepted outright and 27 were conditionally accepted. Almost all conditionally accepted papers were finally accepted (23 of 27), yielding a final acceptance rate of about 32 percent. In addition to the topics that traditionally are represented at machine learning conferences, ICML-2003 also contains papers on applications of machine learning, including exploratory research that describes novel learning tasks; applications that require nonstandard techniques or shed light on limitations of existing learning techniques; and work that investigates the effect of the developers' decisions about problem formulation, representation or data quality on the learning process.