“Participating in the ACTNext Fellow program has been a very enriching experience. It is like working at a front line university, without classrooms. There are interesting conferences, productive scientific discussions, and the people are “around” to interact and collaborate, even more than in the academy. ” – Mariana Cúri

ACTNext was honored to have Dr. Mariana Cúri as our second participant in the ACTNext Fellow program. Mariana is spending a year at the University of Iowa and ACT as part of her sabbatical to research the application of artificial intelligence and machine learning techniques to psychometric models. With the advanced psychometrics group at ACTNext, Mariana is studying the use of variational autoencoders, an unsupervised machine learning algorithm, to fit both cognitive diagnostic models (CDM) and multidimensional item response theory (MIRT) models. Advances in this domain would not only provide a computationally fast and efficient procedure for a notoriously computationally difficult problem, but also provides a machine learning solution to test score equating and missing data imputation.

Biography

Mariana Cúri is currently assistant professor at the Department of Applied Mathematics and Statistics of the University of São Paulo, Brazil. She received the B.Sc., M.Sc., and Ph.D. degrees in Statistics, all of them from the University of São Paulo. Her research interests are mostly in latent variable models, especially in Item Response Theory, Cognitive Diagnostic Models, and Computerized Adaptive Testing.​

About the Program

The ACTNext Fellow program offers individuals from around the country and across the globe the opportunity to come and work with ACTNext scientists in a completely immersive experience. The ACTNext Fellow program was initiated to foster collaboration between ACTNext and other top tier research institutions by providing a rich environment for visiting scientists and scholars to bring their best ideas to bear on both existing ACTNext work, as well as new, joint endeavors. In short, the program aims to put the best minds available in the same room and move the needle on some of today’s most challenging problems facing learning, assessment, and education and career navigation. Program Fellows across diverse fields, from anthropology to data visualization will learn how their own work can benefit from ACTNext’s unique, data-driven approach –  computational psychometrics. This distinctive toolkit blends components from artificial intelligence and machine learning alongside elements from the learning sciences and cutting edge theoretical work in psychometric research.