News:

About Me:

I am a Professor in the Machine Learning Department (in the School of Computer Science) and the Heinz College of Information Systems and Public Policy at Carnegie Mellon University. I’m interested in developing and using computational and data analysis/machine learning methods for solving high impact social good and public policy problems in a fair and equitable way in areas such as criminal justice, education, healthcare, energy, transportation, economic development, workforce development and public safety

Research Expertise: Machine Learning, Public Policy, Social Media Analytics, Text Analytics, Natural Language Processing, Social Networks.

Recently (and reluctantly) added buzzwords: Big Data, Data Science, Artificial Intelligence
Older buzzwords that are trendy now: Machine Learning
Not so old buzzwords that are not trendy now: Data Mining, Analytics

I used to be the Founder/Director of the Center for Data Science and Public Policy, Research Associate Professor in the Department of Computer Science, and a Senior Fellow at the Harris School of Public Policy at the University of Chicago.

What I used to do: Chief Scientist at Obama for America 2012 campaign focusing on analytics, technology, and data. Senior Research Scientist and Director of Analytics research at Accenture Labs where I led a technology research team focused on applied R&D in analytics, machine learning, and data mining for large-scale & emerging business problems in various industries including healthcare, retail & CPG, manufacturing, intelligence, and financial services.

In my ample free time, I advise several analytics start-ups and non-profits, speak at, organize and participate in academic and industry analytics conferences, and publish in machine learning, AI, data science, and Public Policy conferences and journals.

[Longer CV]

Research Interests:

My current work is focused on the question “How do we build AI-Human collaborative systems for social and policy problems that can be trusted to achieve fair and equitable policy outcomes?”

The research pillars include:

  1. Building AI-Human Collaborative Systems (for social and policy problems): How do we use HCI and design approaches to build these systems? How do we make AI systems interpretable and explainable to users to help them improve their decision-making? How do we incorporate feedback?
  2. Ensuring that the outcomes are fair and equitable: How do we define equity? How do we measure and detect bias? How do we mitigate bias to create fair and equitable outcomes?
  3. Making AI systems (for social and policy problems) robust, trustworthy, and resilient?: What is the appropriate validation methodology? How do we make the the system robust to changes over time? How do we make it resilient to attacks? How do we create transparency for different stakeholders?

This work is done through three types of activities:

  1. Training: students and professionals in governments and non profits in the use of data driven methods (Machine Learning, AI, Data Science) for policy and social good
  2. Collaborative Applied Projects: with governments and non profits to solve problems in policy and social good
  3. Research, Tools, and Methodology Development: to develop new methods and tools that are needed across policy areas with special emphasis on increasing fairness, reducing bias, making machine learning and AI algorithms and models more understandable and transparent.

Our areas of interest span health, criminal justice, education, public safety, workforce development, sustainability, transportation, social services, and economic development and we work closely with government agencies (local, federal, international) and NGOs.

If you’re interested in working with me, I’m currently working on the following types of research problems:

  1. Machine Learning methods specifically targeted at the needs of social good and public policy problems.
  2. Challenges in building and deploying Machine learning/AI based systems.
  3. Building explainable and interpretable models for use in human in the loop systems.
  4. Dealing with bias and fairness to create AI systems that result in equitable outcomes.

Historically, my research has spanned from general machine learning and data science to privacy preserving data analysis, text analysis, semi-supervised learning, active learning, information retrieval, Natural Language Processing, and knowledge management. Most of my work has focused on developing and using machine learning & data mining approaches to solve large-scale problems in corporate, political, and non-profit areas.

My current interests lie at the intersection of Machine Learning, Public Policy, and Social Sciences. I’m interested in solving large-scale and high impact social problems using data driven and evidence based methods. A lot of government, civic, and non-profit organizations are realizing the value of better data and have been focusing on improving data collection and data standardization. My goal is to build on these efforts, and work with these organizations to use this data to help improve outcomes in a fair and equitable manner. My work involves developing and using machine learning and social science methods that can be operationalized to solve policy and social challenges across health, criminal justice, education, public safety, social services, and economic development.

Current Projects:

My current projects include:

  • Training
  • Projects
    • working with police departments to build Early Intervention Systems that can help identify officers (and dispatches) at risk of adverse interactions with the public
    • working with public health departments to do preventative lead inspections and reduce lead poisoning in children
    • working with cities to improve home inspection processes to for early blight detection (and prevention) and/or to improve code compliance and health and safety violations in rental properties.
    • working with jurisdictions to combine criminal justice, homelessness, mental health, and medical data to identify people in need of social services and public assistance programs and prevent future incarceration
    • working with school districts to identify students in need of extra support to achieve different educational outcomes (high school graduation for example)
  • Research
    • comparing existing and developing new methods to make machine learning ad artificial intelligence methods and models more understandable and interpretable for use in policy problems.
    • developing bias and fairness audit tools for machine learning models (See Aequitas, our new Bias and Fairness Auidit Tool)
    • conducting a bias and fairness analysis across problems in education, health, criminal justice, and social services.