Oct 23, 2023
Monday
|
09:15 AM - 09:30 AM
|
|
Welcome
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
09:30 AM - 10:30 AM
|
|
Probability Spaces Driven by Geometric Constraints
Wesley Pegden (Carnegie Mellon University)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
What can we understand about probability spaces on "nice" partitions of a geometric region? Can we design efficient samplers? Can we at least detect extreme outliers? These questions have become particularly salient in the past several years as the techniques developed by mathematicians are now applied to conduct statistical analyses of things like U.S. political districtings. We will discuss some recent developments on probability spaces defined by geometric constraints, including positive and negative results on the mixing times of relevant Markov chains, Markov chain methods which eschew mixing-time requirements, and direct sampling methods.
- Supplements
-
--
|
10:30 AM - 11:00 AM
|
|
Break
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
11:00 AM - 12:00 PM
|
|
When Personalization Harms Performance
Berk Ustun (University of California, San Diego)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Machine learning models often include group attributes like sex, age, and HIV status for the sake of personalization – i.e., to assign more accurate predictions to heterogeneous subpopulations. In this talk, I will describe how such practices inadvertently lead to worsenalization, by assigning unnecessarily inaccurate predictions to minority groups. I will discuss how these effects violate our basic expectations from personalization and describe how these violations arise due to standard practices in model development. I will end by highlighting recent work on how to address these issues in practice – first, by setting "personalization budgets" to test for worsenalization; second, by developing models where individuals can consent to personalization at prediction time.
- Supplements
-
--
|
12:00 PM - 01:30 PM
|
|
Lunch
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
01:30 PM - 02:30 PM
|
|
Characterizing General Top Trading Cycles Mechanisms
Bettina Klaus (University of Lausanne)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
In many applied matching problems, indivisible goods that are in unit demand have to be assigned without monetary transfers. One of the most prominent such problems is modeled by classical Shapley-Scarf housing markets (Shapley and Scarf, 1974). Shapley and Scarf (1974) consider an exchange economy in which each agent owns an indivisible object (say, a house); each agent has preferences over houses and wishes to consume exactly one house. The objective of the market designer then is to reallocate houses among agents. When preferences are strict, Shapley and Scarf (1974) show that the strict core (defined by a weak blocking notion) has remarkable features: it is non-empty, and can be easily calculated by the so-called top-trading-cycles (TTC) algorithm (due to David Gale). Moreover, the TTC mechanism that assigns the unique strict core allocation satisfies important incentive properties, strategy-proofness (Roth, 1982) as well as the stronger property of group strategy-proofness (Bird, 1984). Furthermore, Ma (1994) and Svensson (1999) show that the TTC mechanism is the unique mechanism satisfying Pareto efficiency, individual rationality, and strategy-proofness.
After giving a short survey over the wonderful properties the top-trading cycles mechanism has, I’ll consider an extension of Shapley-Scarf housing markets to object allocation problems with coalitional endowments (housing markets with existing tenants are an example). For this relatively new class of problems, I present recent results obtained together with Di Feng during our current stay at the Simons Laufer Mathematical Sciences Institute (formerly MSRI). Our main result is the characterization of sequential priorities-augmented top trading cycles mechanisms.
- Supplements
-
|
03:00 PM - 03:30 PM
|
|
Afternoon Tea
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
|
Oct 24, 2023
Tuesday
|
09:30 AM - 10:30 AM
|
|
Bringing Causality into Fairness: Application to Pretrial Public Safety Assessment
Kosuke Imai
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Using the concept of principal stratification from the causal inference literature, we introduce a new notion of fairness, called principal fairness, for human and algorithmic decision-making. Principal fairness states that one should not discriminate among individuals who would be similarly affected by the decision. Unlike the existing statistical definitions of fairness, principal fairness explicitly accounts for the fact that individuals can be impacted by the decision. We also explain how principal fairness relates to the existing causality-based fairness criteria. In contrast to the counterfactual fairness criteria, for example, principal fairness considers the effects of the decision in question rather than those of protected attributes of interest. Finally, we apply the proposed methodology to preliminary data from the first-ever randomized controlled trial that evaluates the pretrial Public Safety Assessment (PSA) in the criminal justice system.
- Supplements
-
--
|
10:30 AM - 11:00 AM
|
|
Break
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
11:00 AM - 12:00 PM
|
|
Multiplicity in Machine Learning
Flavio Calmon (Harvard University)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
--
- Abstract
This talk reviews the concept of predictive multiplicity in machine learning. Predictive multiplicity arises when different classifiers achieve similar average performance for a specific learning task yet produce conflicting predictions for individual samples. We discuss a metric called “Rashomon Capacity” for quantifying predictive multiplicity in multi-class classification. We also present recent findings on the multiplicity cost of differentially private training methods and group fairness interventions in machine learning.
This talk is based on work published at ICML'20, NeurIPS'22, ACM FAccT'23, and NeurIPS'23.
- Supplements
-
--
|
12:00 PM - 01:30 PM
|
|
Lunch
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
01:30 PM - 02:30 PM
|
|
Fairness in Algorithmic Decisions via Social Choice
Nisarg Shah (University of Toronto)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
As algorithms and AI models are increasingly used to augment, or even replace, traditional human decision-making, there is a growing interest in ensuring that they treat (groups of) people fairly. While fairness is a relatively new design criterion in many areas of algorithmic decision-making (e.g., machine learning), it has a long history of study in social choice theory from microeconomics. In this talk, I will first survey some of the recent advances that boost fairness guarantees in traditional economic problems such as resource allocation, and then show how these can be adapted to many other decision-making paradigms ranging from classification and clustering to recommender systems and conference peer review.
- Supplements
-
--
|
03:00 PM - 03:30 PM
|
|
Afternoon Tea
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
03:30 PM - 04:30 PM
|
|
Best of Both Worlds Fairness
HARIS AZIZ (University of New South Wales)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Best of both worlds fairness is a paradigm in which the goal is to design randomised algorithms that simultaneously achieve desirable fairness properties ex-post and ex-ante. In this talk, I will discuss our results on best of both worlds fairness in various contexts including resource allocation and committee voting.
- Supplements
-
--
|
04:30 PM - 06:20 AM
|
|
Reception
|
- Location
- SLMath: Front Courtyard
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
|
Oct 25, 2023
Wednesday
|
09:30 AM - 10:30 AM
|
|
Online Learning and Collusion in Multi-Unit Auctions
Simina Branzei (Purdue University)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
--
- Abstract
In a carbon auction, licenses for CO2 emissions are allocated among multiple interested players. Inspired by this setting, we consider repeated multi-unit auctions with uniform pricing, which are widely used in practice. Our contribution is to analyze the bidding strategies and properties of these auctions in both the offline and online settings. We also analyze the quality of the equilibria through the lens of the core solution concept. Based on joint work with Mahsa Derakhshan, Negin Golrezaei, and Yanjun Han.
- Supplements
-
Slides
7.49 MB application/pdf
|
|
10:30 AM - 11:00 AM
|
|
Break
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
11:00 AM - 12:00 PM
|
|
Fair Division of Indivisibles
Ruta Mehta (University of Illinois at Urbana-Champaign)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Fair division is the problem of dividing a set of items among $n$ agents in a fair manner. When the items are indivisible, the classical fairness solution concepts of {\em envy-freeness} and {\em proportionality} are rendered inapplicable. In this talk, I will survey recent advances on some of the strongest relaxations of these two concepts, namely EFX and MMS/APS, and their connections to other areas of theory CS, such as extremal combinatorics, probabilistic methods, and the Santa Clause problem.
- Supplements
-
--
|
12:00 PM - 01:30 PM
|
|
Lunch
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
03:00 PM - 03:30 PM
|
|
Afternoon Tea
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
|
Oct 26, 2023
Thursday
|
09:30 AM - 10:30 AM
|
|
Sampling, Optimization, and Evaluating Tradeoffs in Redistricting
Daryl DeFord (Washington State University)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Tools from discrete sampling and optimization have become increasingly important for analyzing graph-based formulations of political redistricting, requiring both operationalizing legislative text and exploring complex Pareto frontiers. In this talk I will discuss recent applications and extensions of these techniques, including for court cases and line-drawing support, evaluating nonpartisan justifications for proposed plans, and balancing multiple population constraints to address within-cycle vote dilution. Along the way I will present related open problems and some proposals based on the cycle basis walk.
- Supplements
-
--
|
10:30 AM - 10:35 AM
|
|
Group Photo
|
- Location
- SLMath: Front Courtyard
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
10:35 AM - 11:00 AM
|
|
Break
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
11:00 AM - 12:00 PM
|
|
Uncovering the Impact of Policy on Redistricting
Gregory Herschlag (Duke University)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
Gerrymandering is the manipulation of political districts to advantage or oppress a particular group. To understand whether districts have been egregiously manipulated, one must obtain baseline outcomes in the absence of manipulation. There is a growing consensus to establish such baselines by sampling a representative policy-based collection of alternative and neutral redistricting plans. Sampling the space of redistricting plans may be recast as sampling a space of graph partitions on a (mostly) planar graph over some family of probability distributions.
Although several research groups have made a number of compelling advances in sampling, there remains a wide chasm between the distributions we are able to sample and the distributions we would like to sample, i.e. we can understand the typical behavior of some policies, but not others. In this talk, I will discuss the redistricting problem along with several novel sampling techniques developed by our research group. These techniques expand the family of distributions (and policies) that we can efficiently sample and include both multi-scale methods and work in parallel tempering.
- Supplements
-
--
|
12:00 PM - 01:30 PM
|
|
Lunch
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
01:30 PM - 02:30 PM
|
|
Open Problem Session
Dana Randall (Georgia Institute of Technology)
|
- Location
- SLMath: Eisenbud Auditorium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
03:00 PM - 03:30 PM
|
|
Afternoon Tea
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
05:30 PM - 08:00 PM
|
|
Social Meeting Gathering
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
|
Oct 27, 2023
Friday
|
09:30 AM - 10:30 AM
|
|
A Heterogeneous Schelling Model for Wealth Disparity and its Effect on Segregation
Dana Randall (Georgia Institute of Technology)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
--
- Abstract
The Schelling model of segregation was introduced in economics to show how micromotives can drive macrobehavior. Agents on a lattice have two colors and try to move to a different location if the number of their neighbors with a different color exceeds some threshold. Simulations reveal that even such mild local color preferences, or homophily, are sufficient to cause segregation. We propose a stochastic generalization of the Schelling model, based on both race and wealth, to understand how carefully architected placement of incentives, such as urban infrastructure, might enhance or mitigate segregation over time.
- Supplements
-
--
|
10:30 AM - 11:00 AM
|
|
Break
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
11:00 AM - 12:00 PM
|
|
Patience Ensures Fairness
Florian Brandl (University of Bonn)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
- Abstract
- --
- Supplements
-
--
|
12:00 PM - 01:30 PM
|
|
Lunch
|
- Location
- --
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
01:30 PM - 03:00 PM
|
|
Panel
Ranya Ahmed (American Civil Liberties Union), Luca Belli (National Institute of Standards and Technology), Stanton Jones (Arnold & Porter), Brooke Madubuonwu (American Civil Liberties Union)
|
- Location
- SLMath: Eisenbud Auditorium, Online/Virtual
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
03:00 PM - 03:30 PM
|
|
Afternoon Tea
|
- Location
- SLMath: Atrium
- Video
-
--
- Abstract
- --
- Supplements
-
--
|
|