Fairness through Optimization

We propose optimization as a general paradigm for formalizing fairness in AI-based decision models. We argue that optimization models allow formulation of a wide range of fairness criteria as social welfare functions, while enabling AI to take advantage of highly advanced solution technology. We show how optimization models can assist fairness-oriented decision making in the context of neural networks, support vector machines, and rule-based systems by maximizing a social welfare function subject to appropriate constraints. In particular, we state tractable optimization models for a variety of functions that measure fairness or a combination of fairness and efficiency. These include several inequality metrics, Rawlsian criteria, the McLoone and Hoover indices, alpha fairness, the Nash and Kalai-Smorodinsky bargaining solutions, combinations of Rawlsian and utilitarian criteria, and statistical bias measures. All of these models can be efficiently solved by linear programming, mixed integer/linear programming, or (in two cases) specialized convex programming methods.

Online Convex Optimization Perspective for Learning from Dynamically Revealed Preferences

We study the problem of online learning (OL) from revealed preferences: a learner wishes to learn an agent's private utility function through observing the agent's utility-maximizing actions in a changing environment. We adopt an online inverse optimization setup, where the learner observes a stream of agent's actions in an online fashion and the learning performance is measured by regret associated with a loss function. Due to the inverse optimization component, attaining or proving convexity is difficult for all of the usual loss functions in the literature. We address this challenge by designing a new loss function that is convex under relatively mild assumptions. Moreover, we establish that the regret with respect to our new loss function also bounds the regret with respect to all other usual loss functions. This then allows us to design a flexible OL framework that enables a unified treatment of loss functions and supports a variety of online convex optimization algorithms. We demonstrate with theoretical and empirical evidence that our framework based on the new loss function (in particular online Mirror Descent) has significant advantages in terms of eliminating technical assumptions as well as regret performance and solution time over other OL algorithms from the literature.

Balancing Fairness and Efficiency in an Optimization Model

Optimization models generally aim for efficiency by maximizing total benefit or minimizing cost. Yet a trade-off between fairness and efficiency is an important element of many practical decisions. We propose a principled and practical method for balancing these two criteria in an optimization model. Following a critical assessment of existing schemes, we define a set of social welfare functions (SWFs) that combine Rawlsian leximax fairness and utilitarianism and overcome some of the weaknesses of previous approaches. In particular, we regulate the equity/efficiency trade-off with a single parameter that has a meaningful interpretation in practical contexts. We formulate the SWFs using mixed integer constraints and sequentially maximize them subject to constraints that define the problem at hand. After providing practical step-by-step instructions for implementation, we demonstrate the method on problems of realistic size involving healthcare resource allocation and disaster preparation. The solution times are modest, ranging from a fraction of a second to 18 seconds for a given value of the trade-off parameter.