Sarah Dean


sarahdean AT eecs DOT berkeley DOT edu

Sarah Dean

I am a PhD student in EECS at UC Berkeley, working with Ben Recht and affiliated with BAIR and BCCI. I am funded by a Berkeley fellowship and the NSF graduate research fellowship.* I am broadly interested in designing methods for and understanding the implications of data-driven optimization, and I have worked specifically in optimal control, machine learning, and computational imaging. My work is motivated by applications in robotics, developmental economics, and recommender systems. I am a founding member of Graduates for Engaged and Extended Scholarship in computing and Engineering (GEESE). I interned with Canopy in Boston, MA during Summer 2019.

I graduated from the University of Pennsylvania in 2016, where I studied electrical engineering and math and had the pleasure of working with professors Daniel Lee and Daniel Koditschek. During my time at Penn, I worked as a teaching assistant for several math and engineering courses and became engaged in service learning through my involvement with the West Philadelphia Tutoring Project at Civic House. I grew up in upstate New York, and I like to spend my time pretending to be athletic outside: hiking, biking, swimming, sailing, roller skating, and cross country skiing.

Research

For the most up to date list of my publications and preprints, see my Google Scholar profile.

Learning for Feedback Control

Guaranteeing Safety of Learned Perception Modules via Measurement-Robust Control Barrier Functions
Sarah Dean, Andrew Taylor, Ryan Cosner, Benjamin Recht, and Aaron Ames.
under review.

Certainty-Equivalent Perception-Based Control [arXiv] [github]
Sarah Dean and Benjamin Recht
under review.

Robust Guarantees for Perception-Based Control [arXiv] [slides] [talk] [poster] [github]
Sarah Dean, Nikolai Matni, Benjamin Recht, and Vickie Ye
presented at L4DC 2020.

On the Sample Complexity of the Linear Quadratic Regulator [arXiv] [FoCM] [talk]
Sarah Dean, Horia Mania, Nikolai Matni, Benjamin Recht, and Stephen Tu
published in Foundations of Computational Mathematics (2019).

Safely Learning to Control the Constrained Linear Quadratic Regulator [arXiv] [slides]
Sarah Dean, Stephen Tu, Nikolai Matni, and Benjamin Recht
presented at ACC 2019.

Regret Bounds for Robust Adaptive Control of the Linear Quadratic Regulator [arXiv] [github]
Sarah Dean, Horia Mania, Nikolai Matni, Benjamin Recht, and Stephen Tu
presented at NeurIPS 2018.

Outcome-Aware Machine Learning

Designing Recommender Systems with Reachability in Mind [PDF] [video]
Sarah Dean, Mihaela Curmei, and Benjamin Recht
presented at Participatory Approaches to Machine Learning workshop at ICML 2020.

Balancing Competing Objectives with Noisy Data: Score-Based Classifiers for Welfare-Aware Machine Learning [arXiv] [github ]
Esther Rolf, Max Simchowitz, Sarah Dean, Lydia T Liu, Daniel Björkegren, Moritz Hardt, and Joshua Blumenstock
presented at ICML 2020.
workshop paper awarded Best Paper at NeurIPS Joint Workshop on AI for Social Good 2019.

Recommendations and User Agency: The Reachability of Collaboratively-Filtered Information [arXiv] [talk]
Sarah Dean, Sarah Rich, and Benjamin Recht
presented at FAT* 2020.

Delayed Impact of Fair Machine Learning [arXiv] [Bloomberg] [BAIR Blog]
Lydia T. Liu, Sarah Dean, Esther Rolf, Max Simchowitz, and Moritz Hardt
Best Paper Award at ICML 2018.

A Broader View on Bias in Automated Decision-Making: Reflecting on Epistemology and Dynamics [arXiv]
Roel Dobbe, Sarah Dean, Thomas Gilbert, and Nitin Kohli
presented at FAT/ML 2018.

Computational Imaging

High-throughput fluorescence microscopy using multi-frame motion deblurring [BOE]
Zachary Phillips, Sarah Dean, Laura Waller, and Benjamin Recht
published in Biomedical Optics Express 11 (2020),
extended abstract awarded Best Student Paper in Imaging Systems at OSA Congress 2018.

*I have made my application available for reference here.

a woman among giants
Last updated 1 September 2020.