Ashia C. Wilson
Ashia C. Wilson
I am a Lister Brothers Career Development Assistant Professor at MIT. My research focuses on designing scalable, reliable and socially responsible AI systems using tools from dynamical systems theory, statistics, and optimization. Here is a list of some information about me (CV, Publications, Contact). Short Bio: I obtained my B.A. from Harvard with a concentration in applied mathematics and a minor in philosophy, and my Ph.D. from UC Berkeley in statistics. Before joining MIT, I held a postdoctoral position in the machine learning and statistics group at Microsoft Research.
Affiliations
Teaching
-
6.401 - Introduction to Data Science and Statistics (Spring 2021)
-
6.036 - Introduction to Machine Learning (Spring 2022, Spring 2023)
-
6.3950 - AI, Decision Making and Society (Fall 2022, Fall 2023, Fall 2024 (syllabus))
Recent Publications & Working Papers
Efficiency of Optimization
-
Adaptive Backtracking for Faster Optimization
Joint with Laurent Lessard and Joao Cavalcanti. 2024.
[arXiv]
-
Fast sampling from constrained spaces using the Metropolis-adjusted Mirror Langevin Algorithm
Joint with Vishwak Srinivasan and Andre Wibisono
Conference on Learning Theory (COLT). 2024.
[arXiv]
-
Mean-field underdamped Langevin dynamics and its spacetime discretization
Joint with Qiang Fu
International Conference on Machine Learning (ICML). 2024.
[arXiv]
-
Accelerated Stochastic Optimization Methods under Quasar-convexity
Joint with Qiang Fu and Dongchu Xu
International Conference on Machine Learning (ICML). 2023.
-
Sufficient conditions for non-asymptotic convergence of Riemannian optimisation methods
Joint with Vishwak Srinivasan
OPTML Workshop, Neurips (Spotlight). 2022.
[arXiv]
Integrity of Predictions
-
Unstable Unlearning: The Hidden Risk of Concept Resurgence in Diffusion Models
Joint with Vinith Suriyakumar, Rohan Alur, Ayush Sekhari and Manish Raghavan. 2024.
[arXiv]
-
Faster Machine Unlearning via Natural Gradient Descent
Joint with Omri Lev. 2024.
[arXiv]
-
The Revealed Preferences of Pre-authorized Licenses and Their Ethical Implications for Generative Models
Joint with Vinith Suriyakumar, Peter Menell and Dylan Hadfield-Menell
The 2nd Workshop on Generative AI and Law (GenLaw). 2024.
[link]
-
Algorithms that Approximate Data Removal: New Results and Limitations
Joint with Vinith Suriyakumar
Advances in Neural Information Processing Systems (NeurIPS). 2022.
[arXiv, code]
Evaluation of Decisions
-
Algorithmic Pluralism: A Structural Approach To Equal Opportunity
Joint with Shomik Jain, Vinith Suriyakumar, and Kathleen Creel
ACM Conference on Fairness, Accountability and Transparency (FAccT, best paper). 2024.
ACM conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO, non-archival). 2023.
[arXiv]
-
Homogenous Algorithms Can Reduce Competition in Personalized Pricing
Joint with Nathan Jo, Kathleen Creel, and Manish Raghavan
ICML Workshop on Humans, Algorithmic Decision-Making and Society: Modeling Interactions and Impact. 2024.
-
Scarce Resource Allocations That Rely On Machine Learning Should Be Randomized
Joint with Shomik Jain and Kathleen Creel
International Conference on Machine Learning (ICML). 2024.
[arXiv]
-
As an AI Language Model, "Yes I Would Recommend Calling the Police": Norm Inconsistency in LLM Decision-Making
Joint with Shomik Jain and Dana Calacci
Conference on AI, Ethics and Society (AIES). 2024.
[arXiv]
-
Automating Transparency Mechanisms in the Judicial System Using LLMs: Challenges and Opportunities
Joint with Ishana Shastri, Shomik Jain and Barbara Engelhardt
Conference on AI, Ethics and Society (AIES). 2024.
[arXiv]