Benjamin Bloem-Reddy

I am Assistant Professor of Statistics at the University of British Columbia. I work on problems in statistics and machine learning, with an emphasis on probabilistic approaches. Most of my ongonig (and future) work is motivated by some form of (scientific) knowledge acquisition, and the interrelated roles of models, data, inference, and prediction. Some recent examples:

  • Uses and benefits of symmetry in modeling, inference, and computation.
  • Integrating probability, statistics, and data with scientific models.
  • The role of causality, broadly construed, in learning from observations of and interactions with the world.
I also collaborate with researchers in the sciences on statistical problems arising in their research.

I was a PhD student with Peter Orbanz at Columbia and a postdoc with Yee Whye Teh in the CSML group at the University of Oxford. Before moving to statistics and machine learning, I studied physics at Stanford University and Northwestern University.

Contact: benbr at stat dot ubc dot ca
Office: Department of Statistics, Earth Sciences Building, Room 3168

Announcements

Postdoctoral researcher opening. Chris Maddison and I have a project listed for the CANSSI Distinguished Postdoctoral Fellowship. If you're interested in statistical aspects of modern machine learning methods, with a focus on invariance/symmetry/equivalence, please consider applying. You'll work with my group primarily on theoretical questions, and collaborate with Chris's group on related methods.

  • High-level project description here.
  • Details on the CANSSI DPDF program and application procedure here. (In short, it's a two-step process. Step 1: You apply directly with CANSSI, who perform a preliminary selection of candidates. You must select 2-3 projects of interest. Step 2: Zoom interviews.)
Application deadline is January 15, 2023. Please get in touch if you have questions!

Research group


Current (alphabetical order)

Past (reverse chronological order)

Papers

Pre-prints

  • Indeterminacy in Latent Variable Models: Characterization and Strong Identifiability
    Q. Xi and B. Bloem-Reddy
    [arxiv]

Published

  • Discussion of F. Caron and E. B. Fox, "Sparse graphs using exchangeable random measures"
    B. Bloem-Reddy
    Journal of the Royal Statistical Society, Series B (Statistical Methodology), 79(5)
    [jrss b] [pdf] [slides from discussion at RSS meeting]

In progress (comments welcome)

Workshop contributions

(Most of these have a more fully developed counterpart above)

  • Generalization Bounds for Invariant Neural Networks
    C. Lyle, B. Bloem-Reddy, Y. Gal, M. Kwiatkowska
    NeurIPS 2019 Workshop on Machine Learning with Guarantees

Teaching

Current

  • Spring 2023: STAT 305, Introduction to Statistical Inference
    Course website on Canvas

Upcoming

Spring 2022: STAT 305, STAT 547S: Topics on Symmetry in Statistics and ML

Past courses

  • Fall 2019, 2020, 2021, 2022: STAT 547C, Topics in Probability
    [course website: 2021, 2022]
  • Summer 2020, Spring 2020, Spring 2021: STAT 305, Introduction to Statistical Inference

Notes

  • Exchangeable random partitions and random discrete probability measures: a brief tour guided by the Dirichlet Process
    B. Bloem-Reddy
    Notes for a lecture given to Oxford PhD students (these are a work in progress)
    [pdf]