Adversaries With Incentives: A Strategic Alternative to Adversarial Robustness

Adversarial training aims to defend against *adversaries*: malicious opponents whose sole aim is to harm predictive performance in any way possible - a rather harsh perspective, which we assert results in unnecessarily conservative models. Instead, we propose to model opponents as simply pursuing their own goals, rather than working directly against the classifier. Employing tools from strategic modeling, our approach uses knowledge or beliefs regarding the opponent's possible incentives as inductive bias for learning. Our method of *strategic training* is designed to defend against opponents within an *incentive uncertainty set*: this resorts to adversarial learning when the set is maximal, but offers potential gains when it can be appropriately reduced. We conduct a series of experiments that show how even mild knowledge regarding the adversary's incentives can be useful, and that the degree of potential gains depends on how incentives relate to the structure of the learning task.

Classification Under Strategic Self-Selection

When users stand to gain from certain predictions, they are prone to act strategically to obtain favorable predictive outcomes. Whereas most works on strategic classification consider user actions that manifest as feature modifications, we study a novel setting in which users decide -- in response to the learned classifier -- whether to at all participate (or not). For learning approaches of increasing strategic awareness, we study the effects of self-selection on learning, and the implications of learning on the composition of the self-selected population. We then propose a differentiable framework for learning under self-selective behavior, which can be optimized effectively. We conclude with experiments on real data and simulated behavior that both complement our analysis and demonstrate the utility of our approach.

One-Shot Strategic Classification Under Unknown Costs

A primary goal in strategic classification is to learn decision rules which are robust to strategic input manipulation. Earlier works assume that strategic responses are known; while some recent works address the important challenge of unknown responses, they exclusively study sequential settings which allow multiple model deployments over time. But there are many domains$\unicode{x2014}$particularly in public policy, a common motivating use-case$\unicode{x2014}$where multiple deployments are unrealistic, or where even a single bad round is undesirable. To address this gap, we initiate the study of strategic classification under unknown responses in the one-shot setting, which requires committing to a single classifier once. Focusing on the users' cost function as the source of uncertainty, we begin by proving that for a broad class of costs, even a small mis-estimation of the true cost can entail arbitrarily low accuracy in the worst case. In light of this, we frame the one-shot task as a minimax problem, with the goal of identifying the classifier with the smallest worst-case risk over an uncertainty set of possible costs. Our main contribution is efficient algorithms for both the full-batch and stochastic settings, which we prove converge (offline) to the minimax optimal solution at the dimension-independent rate of $\tilde{\mathcal{O}}(T^{-\frac{1}{2}})$. Our analysis reveals important structure stemming from the strategic nature of user responses, particularly the importance of dual norm regularization with respect to the cost function.

Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias

Recent studies show that instruction tuning and learning from human feedback improve the abilities of large language models (LMs) dramatically. While these tuning methods can make models generate high-quality text, we conjecture that more implicit cognitive biases may arise in these fine-tuned models. Our work provides evidence that these fine-tuned models exhibit biases that were absent or less pronounced in their pretrained predecessors. We examine the extent of this phenomenon in three cognitive biases - the decoy effect, the certainty effect, and the belief bias - all of which are known to influence human decision-making and reasoning. Our findings highlight the presence of these biases in various models, especially those that have undergone instruction tuning, such as Flan-T5, GPT3.5, and GPT4. This research constitutes a step toward comprehending cognitive biases in instruction-tuned LMs, which is crucial for the development of more reliable and unbiased language models.

Delegated Classification

When machine learning is outsourced to a rational agent, conflicts of interest might arise and severely impact predictive performance. In this work, we propose a theoretical framework for incentive-aware delegation of machine learning tasks. We model delegation as a principal-agent game, in which accurate learning can be incentivized by the principal using performance-based contracts. Adapting the economic theory of contract design to this setting, we define budget-optimal contracts and prove they take a simple threshold form under reasonable assumptions. In the binary-action case, the optimality of such contracts is shown to be equivalent to the classic Neyman-Pearson lemma, establishing a formal connection between contract design and statistical hypothesis testing. Empirically, we demonstrate that budget-optimal contracts can be constructed using small-scale data, leveraging recent advances in the study of learning curves and scaling laws. Performance and economic outcomes are evaluated using synthetic and real-world classification tasks.

Decongestion by Representation: Learning to Improve Economic Welfare in Marketplaces

Congestion is a common failure mode of markets, where consumers compete inefficiently on the same subset of goods (e.g., chasing the same small set of properties on a vacation rental platform). The typical economic story is that prices solve this problem by balancing supply and demand in order to decongest the market. But in modern online marketplaces, prices are typically set in a decentralized way by sellers, with the power of a platform limited to controlling representations -- the information made available about products. This motivates the present study of decongestion by representation, where a platform uses this power to learn representations that improve social welfare by reducing congestion. The technical challenge is twofold: relying only on revealed preferences from users' past choices, rather than true valuations; and working with representations that determine which features to reveal and are inherently combinatorial. We tackle both by proposing a differentiable proxy of welfare that can be trained end-to-end on consumer choice data. We provide theory giving sufficient conditions for when decongestion promotes welfare, and present experiments on both synthetic and real data shedding light on our setting and approach.

Causal Strategic Classification: A Tale of Two Shifts

When users can benefit from certain predictive outcomes, they may be prone to act to achieve those outcome, e.g., by strategically modifying their features. The goal in strategic classification is therefore to train predictive models that are robust to such behavior. However, the conventional framework assumes that changing features does not change actual outcomes, which depicts users as "gaming" the system. Here we remove this assumption, and study learning in a causal strategic setting where true outcomes do change. Focusing on accuracy as our primary objective, we show how strategic behavior and causal effects underlie two complementing forms of distribution shift. We characterize these shifts, and propose a learning algorithm that balances between these two forces and over time, and permits end-to-end training. Experiments on synthetic and semi-synthetic data demonstrate the utility of our approach.

Performative Recommendation: Diversifying Content via Strategic Incentives

The primary goal in recommendation is to suggest relevant content to users, but optimizing for accuracy often results in recommendations that lack diversity. To remedy this, conventional approaches such as re-ranking improve diversity by presenting more diverse items. Here we argue that to promote inherent and prolonged diversity, the system must encourage its creation. Towards this, we harness the performative nature of recommendation, and show how learning can incentivize strategic content creators to create diverse content. Our approach relies on a novel form of regularization that anticipates strategic changes to content, and penalizes for content homogeneity. We provide analytic and empirical results that demonstrate when and how diversity can be incentivized, and experimentally demonstrate the utility of our approach on synthetic and semi-synthetic data.

Learning to Take a Break: Sustainable Optimization of Long-Term User Engagement

Optimizing user engagement is a key goal for modern recommendation systems, but blindly pushing users towards increased consumption risks burn-out, churn, or even addictive habits. To promote digital well-being, most platforms now offer a service that periodically prompts users to take a break. These, however, must be set up manually, and so may be suboptimal for both users and the system. In this paper, we propose a framework for optimizing long-term engagement by learning individualized breaking policies. Using Lotka-Volterra dynamics, we model users as acting based on two balancing latent states: drive, and interest -- which must be conserved. We then give an efficient learning algorithm, provide theoretical guarantees, and empirically evaluate its performance on semi-synthetic data.

Strategic Representation

Humans have come to rely on machines for reducing excessive information to manageable representations. But this reliance can be abused -- strategic machines might craft representations that manipulate their users. How can a user make good choices based on strategic representations? We formalize this as a learning problem, and pursue algorithms for decision-making that are robust to manipulation. In our main setting of interest, the system represents attributes of an item to the user, who then decides whether or not to consume. We model this interaction through the lens of strategic classification (Hardt et al. 2016), reversed: the user, who learns, plays first; and the system, which responds, plays second. The system must respond with representations that reveal `nothing but the truth' but need not reveal the entire truth. Thus, the user faces the problem of learning set functions under strategic subset selection, which presents distinct algorithmic and statistical challenges. Our main result is a learning algorithm that minimizes error despite strategic representations, and our theoretical analysis sheds light on the trade-off between learning effort and susceptibility to manipulation.