PPIG 2019 - 30th Annual Workshop
Doctoral Consortium
Challenging users’ perceptions of decision boundaries in machine learning systems (WIP)
Rob Bowman, Luke Church
Abstract: We investigate the design of smart systems that drive the exploration of decision boundaries created by Machine Learning (ML). We are interested in users’ perceptions of decision boundaries, particularly in scenarios that involve the categorisation of non-categorical data. We present a system for colour categorisation using active learning with decision boundary visualisation and functionality for investigating users’ perceptions of the decision boundaries they are teaching the system.