Adaptive Speed Tests
In: KDML (Hrsg.): Proceedings of the German Workshop on Knowledge Discovery and Machine Learning (KDML 2013)
URL des Volltextes:
4. Beiträge in Sammelwerken; Tagungsband/Konferenzbeitrag/Proceedings
The assessment of a person's traits such as ability is a fundamental problem in human sciences. We focus on assessments of traits that can be mea- sured by determining the shortest time limit al- lowing a testee to solve simple repetitive tasks, so-called speed tests. Existing approaches for ad- justing the time limit are either intrinsically non- adaptive or lack theoretical foundation. By con- trast, we propose a mathematically sound frame- work in which latent competency skills are rep- resented by belief distributions on compact inter- vals. The algorithm iteratively computes a new difficulty setting, such that the amount of be- lief that can be updated after feedback has been received is maximized. We provide theoretical analyses and show empirically that our method performs equally well or better than state of the art baselines in a near-realistic scenario.