Harry Bendekgey's portrait

Harry Bendekgey

I am an Assistant Teaching Professor at Tufts Computer Science with a specialization in machine learning and artificial intelligence. In Fall 2026 I am teaching CS136: Statistical Pattern Recognition and co-instructing both CS61: Discrete Math and CS150: Advanced AI.

Before coming to Tufts, I was a Ph.D. candidate at UC Irvine advised by Erik Sudderth. My research focuses on how we can introduce structure and constraints to deep generative models when we have prior knowledge but limited data. My thesis is available online in the University of California Digital Library.

If you want to connect about research or pedagogy, I can be reached at firstname.last@gmail.com.

I love board games and am always excited to talk about them.

Publications

Learning to Infer Fast by Attending to Sparse Temporal Observations
M Motamed* H Bendekgey*, D Sujono, E Sudderth,
Conference on Uncertainty in Artificial Intelligence, 2026. Paper.
Third-order photon correlations extract single-nanocrystal multiexciton properties in solution
JR Horowitz, DB Berkinsky, H Bendekgey, OJ Tye, T Šverko, KE Shulenberger, and MG Bawendi,
Oprics Express, 2025. Paper.
A Systematic Literature Review of Undergraduate Data Science Education Research
M Dogucu, S Demirci, H Bendekgey, FZ Ricci, and CM Medina,
Journal of Statistics and Data Science Education, 2025. Paper. ArXiv link.
Talks and Presentations: Electronic Conference on Teaching Statistics, 2024.
Scaling study of diffusion in dynamic crowded spaces.
H Bendekgey, G Huber, and D Yllanes,
Journal of Physics A: Mathematical and Theoretical, 2024. Paper. ArXiv link.
Talks and Presentations: American Physical Society March Meeting, 2022.
Unbiased Learning of Deep Generative Models with Structured Discrete Representations.
H Bendekgey, G Hope, and E Sudderth,
Conference on Neural Information Processing Systems, 2023. Paper. ArXiv link.
Talks and Presentations: Pomona College Computer Science Colloquium Series.
Scalable & Stable Surrogates for Flexible Classifiers with Fairness Constraints.
H Bendekgey and E Sudderth,
Conference on Neural Information Processing Systems, 2021. Paper.
Talks and Presentations: Southern California Machine Learning Symposium, 2021.