Scalable Learning of Item Response Theory Models

Susanne Frick, Amer Krivosija, Alexander Munteanu
Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:1234-1242, 2024.

Abstract

Item Response Theory (IRT) models aim to assess latent abilities of $n$ examinees along with latent difficulty characteristics of $m$ test items from categorical data that indicates the quality of their corresponding answers. Classical psychometric assessments are based on a relatively small number of examinees and items, say a class of $200$ students solving an exam comprising $10$ problems. More recent global large scale assessments such as PISA, or internet studies, may lead to significantly increased numbers of participants. Additionally, in the context of Machine Learning where algorithms take the role of examinees and data analysis problems take the role of items, both $n$ and $m$ may become very large, challenging the efficiency and scalability of computations. To learn the latent variables in IRT models from large data, we leverage the similarity of these models to logistic regression, which can be approximated accurately using small weighted subsets called coresets. We develop coresets for their use in alternating IRT training algorithms, facilitating scalable learning from large data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v238-frick24a, title = { Scalable Learning of Item Response Theory Models }, author = {Frick, Susanne and Krivosija, Amer and Munteanu, Alexander}, booktitle = {Proceedings of The 27th International Conference on Artificial Intelligence and Statistics}, pages = {1234--1242}, year = {2024}, editor = {Dasgupta, Sanjoy and Mandt, Stephan and Li, Yingzhen}, volume = {238}, series = {Proceedings of Machine Learning Research}, month = {02--04 May}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v238/frick24a/frick24a.pdf}, url = {https://proceedings.mlr.press/v238/frick24a.html}, abstract = { Item Response Theory (IRT) models aim to assess latent abilities of $n$ examinees along with latent difficulty characteristics of $m$ test items from categorical data that indicates the quality of their corresponding answers. Classical psychometric assessments are based on a relatively small number of examinees and items, say a class of $200$ students solving an exam comprising $10$ problems. More recent global large scale assessments such as PISA, or internet studies, may lead to significantly increased numbers of participants. Additionally, in the context of Machine Learning where algorithms take the role of examinees and data analysis problems take the role of items, both $n$ and $m$ may become very large, challenging the efficiency and scalability of computations. To learn the latent variables in IRT models from large data, we leverage the similarity of these models to logistic regression, which can be approximated accurately using small weighted subsets called coresets. We develop coresets for their use in alternating IRT training algorithms, facilitating scalable learning from large data. } }
Endnote
%0 Conference Paper %T Scalable Learning of Item Response Theory Models %A Susanne Frick %A Amer Krivosija %A Alexander Munteanu %B Proceedings of The 27th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2024 %E Sanjoy Dasgupta %E Stephan Mandt %E Yingzhen Li %F pmlr-v238-frick24a %I PMLR %P 1234--1242 %U https://proceedings.mlr.press/v238/frick24a.html %V 238 %X Item Response Theory (IRT) models aim to assess latent abilities of $n$ examinees along with latent difficulty characteristics of $m$ test items from categorical data that indicates the quality of their corresponding answers. Classical psychometric assessments are based on a relatively small number of examinees and items, say a class of $200$ students solving an exam comprising $10$ problems. More recent global large scale assessments such as PISA, or internet studies, may lead to significantly increased numbers of participants. Additionally, in the context of Machine Learning where algorithms take the role of examinees and data analysis problems take the role of items, both $n$ and $m$ may become very large, challenging the efficiency and scalability of computations. To learn the latent variables in IRT models from large data, we leverage the similarity of these models to logistic regression, which can be approximated accurately using small weighted subsets called coresets. We develop coresets for their use in alternating IRT training algorithms, facilitating scalable learning from large data.
APA
Frick, S., Krivosija, A. & Munteanu, A.. (2024). Scalable Learning of Item Response Theory Models . Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 238:1234-1242 Available from https://proceedings.mlr.press/v238/frick24a.html.

Related Material