Conference Presentation: Hyperparameter Optimization for Neural Network-Based IRT in Constrained Settings
April 9 @ 9:45 am - 10:45 am
“Hyperparameter Optimization for Neural Network-Based IRT in Constrained Settings“
Event: NCME Annual Meeting
Date: Thursday, April 9, 9:45 AM ET – 10:45 AM ET
Session Format: Presentation
Presenters:
Seong Eun Hong
Description:
This study investigates hyperparameter optimization in neural network-based IRT estimation. Using simulation with varying item pool sizes, iterations, and tuning strategies, results show fine-tuning improves performance under constrained settings, offering practical guidelines for implementing neural networks in educational assessment with limited item banks and test administrations
