Why should one do a workshop on failures when the primary goal of each scientific investigation is success? Looking at former and recent publications in evolutionary computation, the general impression is indeed that only success stories are told. As everybody who is experienced in experimental or theoretical research knows that failures happen, and are sometimes even necessary steps to success, the question shall be raised if reporting failures does have a positive impact on further research on a similar topic. Most researchers who gave feedback to our call agreed that it does. However, there is currently no methodology or silent consent on which failures to report and how. Furthermore, it is even unclear how much different frequently occurring failures are. As this was to our knowledge the first workshop on this topic in evolutionary computation, the most important task was thus to collect a number of cases and get an overview of the predominant types of failures.
/lp/association-for-computing-machinery/learning-from-failures-in-evolutionary-computation-gecco-2009-events-v6xVEm64Y3