Miruna Oprescu

About

I am a Ph.D. candidate in Computer Science at Cornell University (Cornell Tech) advised by Prof. Nathan Kallus. I am also a Department of Energy Computational Science Graduate Fellow (DOE CSGF). My research focuses on developing robust and reliable machine learning algorithms for causal inference and data-driven decision-making. I am particularly interested in tackling real-world challenges in causal inference applications. Some of the topics I explore in my research include:

Individualized treatment effect estimation from observational data [ICML'19, NeurIPS'19]
Risk and sensitivity analysis in observational causal inference studies [AISTATS'23, ICML'23, NeurIPS'24]
Experimental data with imperfect compliance [NeurIPS'24]
Complex data patterns such as network or spatio-temporal data

If you would like to collaborate on any of these topics and beyond, please reach out!

Bio

Before my PhD, I was a Senior Data and Applied Scientist at Microsoft Research, where I developed data science tools for various applications, including causal inference, cancer research, and weather forecasting. I was a key member of the ALICE team and a core contributor to the EconML Python library for causal inference. During my PhD, I interned at Netflix and Brookhaven National Laboratory. I hold an A.B. in Physics and Mathematics with a minor in Computer Science from Harvard University.

News

[11/20/24] Gave a talk at the Workshop in Operations Research and Data Science (WORDS), Duke University: Reliable Treatment Effect Estimation Using Weak Instruments and Observational Data. Talk abstract.
[09/26/24] Paper Estimating Heterogeneous Treatment Effects by Combining Weak Instruments and Observational Data was accepted at NeurIPS 2024!
[09/26/24] Paper Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes was accepted at NeurIPS 2024!
[08/23/24] Wrapped up my internship at the Brookhaven National Lab, Computational Science Initiative.
[08/14/24] Passed my admission to candidacy exam (A Exam)! Talk slides.
[07/17/24] Presented my research at the 2024 DOE CSGF Annual Program Review in D.C. Poster and photo.
[06/10/24] New Preprint: Estimating Heterogeneous Treatment Effects by Combining Weak Instruments and Observational Data.
[03/29/24] New Preprint: Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes.
[01/19/24] Paper Low-rank MDPs with Continuous Action Spaces was accepted at AISTATS 2024!
[11/06/23] New Preprint: Low-Rank MDPs with Continuous Action Spaces.
[09/27/23] Gave a talk at the Brookhaven National Lab, Computational Science Initiative: Uncertainty Quantification in Causal Inference: Sharp and Efficient Bounds on Heterogeneous Causal Effects Under Hidden Confounding. Talk abstract.
[07/19/23] Presented my research at the 2023 DOE CSGF Annual Program Review in D.C. Poster.
[04/24/23] Paper B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding was accepted at ICML 2023!
[04/20/23] New Preprint: B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding.
[01/20/23] Paper Robust and Agnostic Learning of Conditional Distributional Treatment Effects was accepted at AISTATS 2023!
[08/02/22] Finished my internship at Netflix with the Search and Recommendations team.
New Preprint: Robust and Agnostic Learning of Conditional Distributional Treatment Effects.
[04/06/22] Awarded the prestigious Department of Energy Computational Science Graduate Fellowship for 2022-2026! Announcement.
[02/02/22] Named finalist of the 2022 Meta PhD Research Fellowship. Announcement.

Publications

Most recent publications on Google Scholar.
* Equal contribution. ^‡ Alphabetical order.

Selected
Preprints
All

Estimating Heterogeneous Treatment Effects by Combining Weak Instruments and Observational Data

Miruna Oprescu, Nathan Kallus

NeurIPS'24: Advances in Neural Information Processing Systems. 2024.

Paper Arxiv Slides Code

Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes

Andrew Bennett^‡, Nathan Kallus^‡, Miruna Oprescu^‡, Wen Sun^‡, Kaiwen Wang^‡

NeurIPS'24: Advances in Neural Information Processing Systems. 2024.

Paper Arxiv Code

B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding

Miruna Oprescu, Jacob Dorn, Marah Ghoummaid, Andrew Jesson, Nathan Kallus, Uri Shalit

ICML'23: International Conference on Machine Learning. 2023.

Paper Arxiv Poster Code

Robust and Agnostic Learning of Conditional Distributional Treatment Effects

Nathan Kallus*, Miruna Oprescu*

AISTATS'23: International Conference on Artificial Intelligence and Statistics. 2023.

Paper Arxiv Poster Code

Estimating the Long-Term Effects of Novel Treatments

Keith Battocchi^‡, Eleanor Dillon^‡, Maggie Hei^‡, Greg Lewis^‡, Miruna Oprescu^‡, Vasilis Syrgkanis^‡

NeurIPS'21: Advances in Neural Information Processing Systems. 2021.

Paper Arxiv

Causal Inference and Machine Learning in Practice with EconML and CausalML: Industrial Use Cases at Microsoft, TripAdvisor, Uber

Vasilis Syrgkanis, Greg Lewis, Miruna Oprescu, Maggie Hei, Keith Battocchi, Eleanor Dillon, Jing Pan, Yifeng Wu, Paul Lo, Huigang Chen, Totte Harinen, Jeong-Yoon Lee

SIGKDD'21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (Tutorial). 2021.

Slides

EconML: A Machine Learning Library for Estimating Heterogeneous Treatment Effects

Miruna Oprescu, Vasilis Syrgkanis, Keith Battocchi, Maggie Hei, Greg Lewis

NeurIPS'19 Do the right thing: Machine Learning and Causal Inference for Improved Decision Making. 2019.

Paper Slides Poster Code

Machine Learning Estimation of Heterogeneous Treatment Effects with Instruments

Vasilis Syrgkanis, Victor Lei, Miruna Oprescu, Maggie Hei, Keith Battocchi, Greg Lewis

NeurIPS'19: Advances in Neural Information Processing Systems. 2019. Spotlight.

Paper Arxiv Poster Code

Orthogonal Random Forest for Causal Inference

Miruna Oprescu^‡, Vasilis Syrgkanis^‡, Zhiwei Steven Wu^‡

ICML'19: International Conference on Machine Learning. 2019.

Paper Arxiv Poster Code

Estimating Heterogeneous Treatment Effects by Combining Weak Instruments and Observational Data

Miruna Oprescu, Nathan Kallus

NeurIPS'24: Advances in Neural Information Processing Systems. 2024.

Paper Arxiv Slides Code

Low-rank MDPs with Continuous Action Spaces

Andrew Bennett^‡, Nathan Kallus^‡, Miruna Oprescu^‡

AISTATS'24: International Conference on Artificial Intelligence and Statistics. 2024.

Paper Arxiv Poster

Efficient and Sharp Off-Policy Evaluation in Robust Markov Decision Processes

Andrew Bennett^‡, Nathan Kallus^‡, Miruna Oprescu^‡, Wen Sun^‡, Kaiwen Wang^‡

NeurIPS'24: Advances in Neural Information Processing Systems. 2024.

Paper Arxiv Code

SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking

Soukayna Mouatadid, Paulo Orenstein, Genevieve Flaspohler, Miruna Oprescu, Judah Cohen, Franklyn Wang, Sean Knight, Maria Geogdzhayeva, Sam Levang, Ernest Fraenkel, Lester Mackey

NeurIPS'23: Advances in Neural Information Processing Systems. 2023.

Paper Arxiv Code Data

B-Learner: Quasi-Oracle Bounds on Heterogeneous Causal Effects Under Hidden Confounding

Miruna Oprescu, Jacob Dorn, Marah Ghoummaid, Andrew Jesson, Nathan Kallus, Uri Shalit

ICML'23: International Conference on Machine Learning. 2023.

Paper Arxiv Poster Code

Adaptive Bias Correction for Improved Subseasonal Forecasting

Soukayna Mouatadid, Paulo Orenstein, Genevieve Flaspohler, Judah Cohen, Miruna Oprescu, Ernest Fraenkel, Lester Mackey

Nature Communications. 2023.

Paper Arxiv Code

Robust and Agnostic Learning of Conditional Distributional Treatment Effects

Nathan Kallus*, Miruna Oprescu*

AISTATS'23: International Conference on Artificial Intelligence and Statistics. 2023.

Paper Arxiv Poster Code

Estimating the Long-Term Effects of Novel Treatments

Keith Battocchi^‡, Eleanor Dillon^‡, Maggie Hei^‡, Greg Lewis^‡, Miruna Oprescu^‡, Vasilis Syrgkanis^‡

NeurIPS'21: Advances in Neural Information Processing Systems. 2021.

Paper Arxiv

Causal Inference and Machine Learning in Practice with EconML and CausalML: Industrial Use Cases at Microsoft, TripAdvisor, Uber

Vasilis Syrgkanis, Greg Lewis, Miruna Oprescu, Maggie Hei, Keith Battocchi, Eleanor Dillon, Jing Pan, Yifeng Wu, Paul Lo, Huigang Chen, Totte Harinen, Jeong-Yoon Lee

SIGKDD'21: Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (Tutorial). 2021.

Slides

Online Learning with Optimism and Delay

Genevieve E Flaspohler, Francesco Orabona, Judah Cohen, Soukayna Mouatadid, Miruna Oprescu, Paulo Orenstein, Lester Mackey

ICML'21: International Conference on Machine Learning. 2021.

Paper Arxiv Code

Multifactorial Model to Predict Response to PD-(L) 1 Blockade in Patients with High PD-L1 Metastatic Non-Small Cell Lung Cancer

Kathryn Arbour, Miruna Oprescu, Joe Hakim, Hira Rizvi, Max Leiserson, Mark Ginsburg, Andrew Plodkowski, Jay Sauter, Isabel Preeshagul, Sharon Gillett, Philip Rosenfield, Lester Mackey, Miro Dudik, Matthew Hellmann

Journal of Thoracic Oncology (Extended Abstract). 2019.

Paper Arxiv

EconML: A Machine Learning Library for Estimating Heterogeneous Treatment Effects

Miruna Oprescu, Vasilis Syrgkanis, Keith Battocchi, Maggie Hei, Greg Lewis

NeurIPS'19 Do the right thing: Machine Learning and Causal Inference for Improved Decision Making. 2019.

Paper Slides Poster Code

Machine Learning Estimation of Heterogeneous Treatment Effects with Instruments

Vasilis Syrgkanis, Victor Lei, Miruna Oprescu, Maggie Hei, Keith Battocchi, Greg Lewis

NeurIPS'19: Advances in Neural Information Processing Systems. 2019. Spotlight.

Paper Arxiv Poster Code

Orthogonal Random Forest for Causal Inference

Miruna Oprescu^‡, Vasilis Syrgkanis^‡, Zhiwei Steven Wu^‡

ICML'19: International Conference on Machine Learning. 2019.

Paper Arxiv Poster Code

Vitæ

Full CV in PDF.

Cornell University 2021 - now

Ph.D. in Computer Science
Minor: Operations Research
Adviser: Nathan Kallus
Brookhaven National Laboratory May-Aug 2024

Research Intern
Computational Science Initiative
Mentor: Shinjae Yoo
Cornell University August 2024

M.S. in Computer Science
Awarded enroute to Ph.D.
Netflix May-Aug 2022

Machine Learning Intern
Customer Machine Learning Group
Mentors: Sudeep Das, Aish Fenton
Microsoft Research 2017 - 2021

Senior Data and Applied Scientist
ALICE Team
Core Contributior to EconML
Microsoft 2015 - 2017

Software Development Engineer
Cloud Machine Learning Group
Harvard University 2011 - 2015

Joint A.B. in Physics and Mathematics
Minor: Computer Science
Johns Hopkins University May-Aug 2014

Research Intern
Mentor: Suchi Saria

Website Design

This website was created using Jekyll and is based on a template by Martin Saveski. A huge thank you to Martin for designing this fantastic theme! You can find the code used to build the site in this GitHub repo.