Avatar

Paulo Orenstein

Assistant Professor

IMPA


I'm an assistant professor at IMPA. My research is focused on the interplay between statistics, probability, and computation. On the theory side, I am interested in high-dimensional Bayesian models, Monte Carlo methods and robust mean estimation. On the applications side, I have been using machine learning to extend weather forecasts into the subseasonal realm. I enjoy working on both theoretical and applied projects, and find them to often illuminate each other.

Before coming to IMPA, I obtained a PhD in Statistics at Stanford University, advised by Persi Diaconis, and received a Masters in Mathematics and BSc in Economics, both from PUC-Rio, in Brazil. I also spent a quarter at Google Research and another at Meta Reality Labs.

Interests

  • High-dimensional Bayesian Statistics
  • Robust Mean Estimation
  • Monte Carlo Methods
  • Applied Statistics and Machine Learning

Education

  • PhD in Statistics, 2019

    Stanford University

  • Masters in Mathematics, 2014

    PUC-Rio

  • BSc in Economics, 2012

    PUC-Rio

Recent Papers

Adaptive Bias Correction for Improved Subseasonal Forecasting
SubseasonalClimateUSA: A Dataset for Subseasonal Forecasting and Benchmarking
AmnioML: Amniotic Fluid Segmentation and Volume Prediction with Uncertainty Quantification
Online Learning with Optimism and Delay

Recent Talks

Split Conformal Prediction for Non-Exchangeable Data

Extending split conformal prediction to non-exchangeable data through concentration inequalities and decoupling properties.

Subseasonal Toolkit

Augmenting metereological forecasts with learned statistical corrections yields advanced subseasonal models at low computational cost.

Uncertainty Quantification for Amniotic Fluid Segmentation

Deep learning and conformal prediction algorithms for amniotic fluid segmentation and volume estimation from fetal MRIs.

ExactBoost

ExactBoost is an boosting algorithm tailored to combinatorial and non-decomposable metrics, with provable generalization guarantees.

Forecast Rodeo

Results and award-winning methods of a year-long data challenge to predict the weather 2-6 weeks in advance.

Teaching

Machine Learning
Machine Learning
Projects in Machine Learning (Hashing)

Contact

  • pauloo@impa.br
  • Estrada Dona Castorina 110, Rio de Janeiro, Brazil, 22460-320