Visual Knowledge Tracing

AbstractEach year, thousands of people learn new visual categorization tasks – radiologists to recognize tumors, birdwatchers distinguish similar species, and crowd workers how annotate valuable data for applications like autonomous driving. As humans learn, their brain updates the features it extracts attend to, which ultimately informs final classification decisions. In this work, we propose a novel task tracing evolving behavior human learners as they engage in challenging tasks. We models that jointly extract used by well predicting functions utilize. collect three datasets from real order evaluate performance different knowledge methods. Our results show our recurrent are able predict on medical image species identification tasks.KeywordsVisual classificationKnowledge tracingHuman learning