EquiTensors: Learning Fair Integrations of Urban Mobility Data

UW's Bill Howe: EquiTensors Learning Fair Integrations of Urban Mobility Data

February 22, 2021

Bill HoweBill Howe, University of Washington Associate Professor in the Information School, Adjunct Associate Professor in Computer Science & Engineering, and Associate Director and Senior Data Science Fellow at the UW eScience Institute, presented EquiTensors: Learning Fair Integrations of Urban Mobility Data at the Applications of Data Science and AI to Equity, Race, and Inclusion in Mobility and Transportation seminar series sponsored by the Institute of Transportation Studies, College of Engineering, and Center for Information Technology Research in the Interest of Society and the Banatao Institute on Feb. 19.

This Zoom seminar series "Applications of Data Science and AI to Equity, Race, and Inclusion in Mobility and Transportation" will bring a unique and innovative perspective to existing discussions around diversity, equity, and inclusion. Our aim is to reflect on and raise awareness of applications, opportunities, and potential misuses of data science and AI applied to mobility and transportation, specifically as it refers to race, equity, and diversity.

Abstract: Neural methods are state-of-the-art for urban mobility prediction problems. Just as pre-trained text models are state-of-the-art for natural language processing applications and pre-trained image models are state-of-the-art for computer vision applications, we are exploring pre-trained city models that can be reused across a variety of urban prediction tasks. We consider two key principles: 1) Model performance depends on making use of all available data (e.g., weather, housing prices, traffic, etc.) because everything is interdependent; 2) All city data is polluted by systemic discrimination, and model training will only reinforce that discrimination unless explicitly counteracted.

In this talk, I’ll describe the EquiTensors project, where we aim to learn fair, reusable, integrated representations of heterogeneous city data to improve performance of mobility prediction tasks while managing equity considerations. We first align source datasets to a consistent spatio-temporal domain, then describe a self-supervised model based on convolutional denoising autoencoders to learn shared representations. We extend this core integrative model with adaptive weighting to prevent certain datasets from dominating the signal. To combat discriminatory signals in the data, we use an adversarial model to "unlearn" correlations with a sensitive attribute (e.g., race or income). Experiments with 23 input datasets on multiple mobility applications show that the learned representations EquiTensors can simultaneously improve performance of downstream applications while mitigating discriminatory effects.

I'll end by describing our broader emphasis on "data equity systems." As the deployment of automated decision tools continues to accelerate, their interactions with fundamental questions in law, in the social sciences, and in public policy have become impossible to ignore. The technology holds the promise of reducing costs, reducing errors, and improving objectivity, but there is enormous potential for harm: As we train algorithms on biased data, we are amplifying, operationalizing, and, most insidiously, legitimizing the historical discrimination and opacity that the technology was in part intended to address. We argue that data systems research needs to broaden scope to explicitly model, manage, and communicate assumptions about the contexts in which they are deployed, making equity issues a first-class design consideration.

Bio: Bill Howe is Associate Professor in the Information School and Adjunct Associate Professor in the Allen School of Computer Science & Engineering and the Department of Electrical Engineering. His research interests are in data management, machine learning, and visualization, particularly as applied in the physical and social sciences. As Founding Associate Director of the UW eScience Institute, Dr. Howe played a leadership role in the Moore-Sloan Data Science Environment program through a $32.8 million grant awarded jointly to UW, NYU, and UC Berkeley, and founded UW’s Data Science for Social Good Program. With support from the MacArthur Foundation, NSF, and Microsoft, Howe directs UW’s participation in the Cascadia Urban Analytics Cooperative. He founded the UW Data Science Masters Degree, serving as its inaugural Program Chair, and created a first MOOC on data science that attracted over 200,000 students. His research has been featured in the Economist and Nature News, and he has authored award-winning papers in conferences across data management, machine learning, and visualization. He has a Ph.D. in Computer Science from Portland State University and a Bachelor’s degree in Industrial & Systems Engineering from Georgia Tech.