AI FOR ECONOMICS

SEMINAR CO-ORGANIZED BY HARVARD ECON-CS & SALESFORCE AI RESEARCH
  • Can machine learning be used to help with the development of effective economic policy?
  • Can we understand economic behavior through granular, economic data sets?
  • Can we automate economic transactions for individuals?
  • Can we build rich and faithful simulations of economic systems with strategic agents?

SCHEDULE

UPCOMING: Noon EST, 9 am PST. April 6, 2021

John Dickerson

University of Maryland

Deep Learning for Auction Design: Fairness, Robustness, and Expressiveness

The design of revenue-maximizing auctions with strong incentive guarantees is a core concern of economic theory. Computational auctions enable online advertising, sourcing, spectrum allocation, and myriad financial markets. Analytic progress in this space is notoriously difficult; since Myerson's 1981 work characterizing single-item optimal auctions, there has been limited progress outside of restricted settings. A recent paper by Dütting et al. circumvents analytic difficulties by applying deep learning techniques to, instead, approximate optimal auctions. Their RegretNet architecture can represent auctions with arbitrary numbers of items and participants; it is trained to be empirically strategyproof, but the property is never exactly verified leaving potential loopholes for market participants to exploit. In parallel, new research from Ilvento et al. and other groups has developed notions of fairness in the context of auction design. Inspired by these advances, in this talk, we discuss extensions of these techniques for approximating auctions using deep learning to address concerns of * fairness while maintaining high revenue and strong incentive guarantees; * certified robustness, that is, verification of claimed strategyproofness of deep learned auctions; and * expressiveness via different demand functions and other constraints. To enable that last point, we propose a new architecture to learn incentive compatible, revenue-maximizing auctions from sampled valuations, which uses the Sinkhorn algorithm to perform a differentiable bipartite matching. Our new framework allows the network to learn strategyproof revenue-maximizing mechanisms in settings not learnable by the previous RegretNet architecture. This talk covers hot-off-the-presses work led by PhD students Michael Curry, Ping-yeh Chiang, and Samuel Dooley, and undergraduate students Kevin Kuo, Uro Lyi, Anthony Ostuni, and Elizabeth Horishny. Papers have appeared at NeurIPS-20 or are currently under review; please check arXiv or get in touch for drafts.

John P Dickerson is an Assistant Professor of Computer Science at the University of Maryland as well as Chief Scientist of Arthur AI, an enterprise-focused AI/ML model monitoring firm. He is a recipient of awards such as the NSF CAREER Award, IEEE Intelligent Systems AI's 10 to Watch, Google Faculty Research Award, Google Research Scholar, and paper awards and nominations at venues such as AAAI. His research centers on solving practical economic problems using techniques from computer science, stochastic optimization, and machine learning. He has worked extensively on theoretical and empirical approaches to organ exchange where his work has set policy at the UNOS nationwide kidney exchange; worldwide blood donation markets with Facebook; game-theoretic approaches to counter-terrorism and negotiation, where his models have been deployed; and market design problems in industry (e.g., online advertising) through various startups. Dickerson received his PhD in computer science from Carnegie Mellon University.

FULL SCHEDULE

Date Speaker Title
Oct 6, 2020 Doyne Farmer
(Oxford)
Global Microeconomics
Video
Nov 10, 2020 Amy Greenwald
(Brown)
Learning Equilibria in Simulation-Based Games ... and the Ensuing Empirical Design of Mechanisms
Video
Dec 15, 2020 Lihong Li
(Google)
Estimating Long-term Rewards by Off-policy Reinforcement Learning
Video
Feb 2, 2021 Fei Fang
(CMU)
Game Theory and Machine Learning for Multiagent Communication and Coordination
Video  -  Notes (thanks Yaman Habip!)
March 9, 2021 Thore Graepel
(DeepMind, UCL)
Automatic Curricula in Deep Multi-agent Reinforcement Learning
Video
April 6, 2021 John Dickerson
(University of Maryland)
Deep Learning for Auction Design: Fairness, Robustness, and Expressiveness
Video

TOPICS

Machine learning offers enormous potential to transform our understanding of economics, economic decision making, and public policy. Yet its adoption by economists, social scientists, and policymakers remains nascent.

This seminar series will highlight both the opportunities as well as the barriers to the adoption of ML in economics. In particular, we aim to accelerate the use of machine learning to rapidly develop, test, and deploy effective economic policies that are grounded in representative data.

This seminar series will expose some of the critical socio-economic issues that stand to benefit from applying machine learning, expose underexplored economic datasets and simulations, and identify machine learning research directions that would have a significant positive socio-economic impact. This includes policies and mechanisms that target socio-economic issues such as diversity and fair representation in economic outcomes, economic equality, and improving economic opportunity.

ECONOMICS

  • Inequality and social mobility
  • Sustainability
  • Innovation + entrepreneurship
  • Market design (e.g., labor, capital, consumer-facing)
  • Taxation
  • Behavioral economics
  • Game theory
  • Data-driven policy-making, and collecting representative and robust economic datasets

MACHINE LEARNING

  • Reinforcement learning: multi-agent RL, cooperation, social dilemmas, principal-agent problems, equilibria and solution concepts.
  • Inverse reinforcement learning
  • Transfer from simulation to the real world
  • Multi-objective and constrained optimization
  • Causal inference
  • Explainability
  • Ethical issues: addressing bias in economic data, learning equitable policies, privacy-preserving learning.

ORGANIZATION

David C. Parkes

Harvard

Alex Trott

Salesforce

Stephan Zheng

Salesforce