Amortised Design Optimization for Item Response Theory

Antti Keurulainen*, Isak Westerlund, Oskar Keurulainen, Andrew Howes

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

1 Citation (Scopus)

Abstract

Item Response Theory (IRT) is a well known method for assessing responses from humans in education and psychology. In education, IRT is used to infer student abilities and characteristics of test items from student responses. Interactions with students are expensive, calling for methods that efficiently gather information for inferring student abilities. Methods based on Optimal Experimental Design (OED) are computationally costly, making them inapplicable for interactive applications. In response, we propose incorporating amortised experimental design into IRT. Here, the computational cost is shifted to a precomputing phase by training a Deep Reinforcement Learning (DRL) agent with synthetic data. The agent is trained to select optimally informative test items for the distribution of students, and to conduct amortised inference conditioned on the experiment outcomes. During deployment the agent estimates parameters from data, and suggests the next test item for the student, in close to real-time, by taking into account the history of experiments and outcomes.

Original languageEnglish
Title of host publicationArtificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky - 24th International Conference, AIED 2023, Proceedings
EditorsNing Wang, Genaro Rebolledo-Mendez, Vania Dimitrova, Noboru Matsuda, Olga C. Santos
PublisherSpringer
Pages359-364
Number of pages6
ISBN (Print)978-3-031-36335-1
DOIs
Publication statusPublished - 2023
MoE publication typeA4 Conference publication
EventInternational Conference on Artificial Intelligence in Education - Tokyo, Japan
Duration: 3 Jul 20237 Jul 2023
Conference number: 24

Publication series

NameCommunications in Computer and Information Science
Volume1831 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

ConferenceInternational Conference on Artificial Intelligence in Education
Abbreviated titleAIED
Country/TerritoryJapan
CityTokyo
Period03/07/202307/07/2023

Keywords

  • Deep Reinforcement Learning (DRL)
  • Experimental Design
  • Item Response Theory (IRT)

Fingerprint

Dive into the research topics of 'Amortised Design Optimization for Item Response Theory'. Together they form a unique fingerprint.

Cite this