Overall theme of Rohit’s research

Rohit’s research currently centers on sequential decision-making in high-stakes environments, where statistical (online) inference is critical, and valid uncertainty quantification (UQ) protects automated decision-makers from committing potentially costly errors. The overarching goal of his research is to develop efficient, stable, robust, fair and trustworthy artificial intelligent (AI) systems, capable of making well-reasoned decisions with valid UQ.

Research interests:

• Theory and methods: Sequential learning & inference, experimental design & causal inference, preference learning, predictive inference
• Application areas: Health communication, public health, clinical trial, sports, genetics

Publication:

• Kanrar, R., Jiang, F., Cai, Z. (2025) Model-free Change-point Detection using AUC of a Classifier. Journal of Machine Learning Research {arXiv/Github/Journal}

Preprints:

• Kanrar, R., Li, C., Ghodsi, Z., Gamalo, M. (2025+). Risk-inclusive Contextual Bandits for Early Phase Clinical Trials. Under revision (Round 2) {arXiv/Github/Poster}
• King, A., Liao, Y., Chen, T., Kanrar, R., Chunara, R., Margolin, D., Nettleton, D., Niederdeppe, J. (2024+) Testing the effects of segmented crowdsource-selected messages to improve intentions to follow colorectal cancer screening recommendations: Study protocol for a randomized controlled trial. Under review