Papers

A list of my published papers and preprints in philosophy of machine learning, feminist philosophy, and philosophy of physics.

AI Ethics / Philosophy of Machine Learning

The Disagreement Problem in Faithfulness Metrics”. In the NeurIPS 2023 XAI in Action workshop. Co-authored with Brian Barr, Noah Fatsi, Alden Richter, Caleb Mok, and Danny Proano.

  • Post-hoc explainability methods for black-box machine learning models are sometimes evaluated based on how “faithful” they are to the black-box model. Our paper presents empirical evidence that various faithfulness metrics proposed by researchers actually disagree when they are used to “rank” different explanation methods.

“Making Intelligence: Ethical Values in IQ and ML Benchmarks”. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency (FAccT ‘23). Co-authored with Borhane Blili Hamelin.

  • Uses lessons from feminist and anti-racist critiques of IQ benchmarks to point to similar risks in how we create machine learning benchmarks.
  • Journal link, Preprint

“Should attention be all we need? The epistemic and ethical implications of unification in machine learning.” Forthcoming in Conference on Fairness, Accountability, and Transparency (FAccT ‘22). Co-authored with Nic Fishman.

  • Assesses the epistemic and ethical risks/benefits of the trend towards unified model architectures in machine learning.
  • Journal Link, Preprint

“Epistemic values in feature importance methods: Lessons from feminist epistemology”. Conference on Fairness, Accountability, and Transparency (FAccT ‘21). Co-authored with Lizzie Kumar.

  • Argues that many popular feature importance methods in ML adhere to a methodology and view of objectivity that is in tension with feminist epistemology.
  • Winner of FAccT 2021 best paper award.
  • Journal link, preprint

Beyond Methods Reproducibility in Machine Learning”. ML-Retrospectives, Surveys & Meta-Analyses Workshop at NeurIPS 2020.

  • Argues that reproducibility definitions in ML are too focused on methods reproducibility, when in fact we need different types of reproducibilities for different types of claims made by ML experiments.

“Robustness in Machine Learning Explanations: Does It Matter?” Conference on Fairness, Accountability, and Transparency (FAccT ’20), January 27–30, 2020, Barcelona, Spain. ACM, New York, NY, USA.

  • Applies the concept of derivational robustness to explanations generated for black-box machine learning models.
  • Journal link, preprint

Philosophy of Physics

“Solutions in Constructive Field Theory,” Philosophy of Science 84, no. 2 (April 2017): 335-358.

  • On how the concept of a “correct solution” in mathematically rigorous quantum field theory depends on non-rigorous considerations.
  • Journal link; preprint

“Coarse-Graining as a Route to Microscopic Physics: The Renormalization Group in Quantum Field Theory,” Philosophy of Science 82, no. 5 (December 2015): 1211-1223.

  • Why the renormalization group matters even if we care about removing all cutoffs from quantum field theory.
  • Winner of the Philosophy of Science Association’s Recent PhD Essay Award.
  • Journal Link, preprint

“Interpretive strategies for deductively insecure theories: The case of early quantum electrodynamics,” Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics, Volume 44, Issue 4 (2013): 395-403.

  • Historically interpreting early quantum electrodynamics through the lens of Bill Wimsatt’s work on false models. Reading early QED in German was fun!
  • Journal link, preprint

Feminist Philosophy

“Idealization and Abstraction in Models of Injustice”, Hypatia, 32 (2017): 329-346.

  • On whether feminist philosophers should eschew using idealized models in social and political theory.
  • Journal Link, preprint

“Defining the Terms of Our Struggle: On the Role of the Concept of Woman in Political Theory”

  • Yet another take on how we should define “woman” for the purposes of social justice. Attempts to be more trans-friendly than some other takes on the issue.
  • Unpublished manuscript.