Explanatory Debugging: Supporting End-User Debugging of Machine-Learned Programs

摘要: Many machine-learning algorithms learn rules of behavior from individual end users, such as task-oriented desktop organizers and handwriting recognizers. These form a “program” that tells the computer what to do when future inputs arrive. Little research has explored how an user can debug these programs they make mistakes. We present our progress toward enabling users learned via Natural Programming methodology. began with formative study exploring reason about correct text-classification program. From results, we derived prototyped concept based on “explanatory debugging”, then empirically evaluated it. Our results contribute methods for exposing program’s logic eliciting corrections improve predictions.