Automated bidding to optimize online advertising with various constraints, e.g. ROI constraints and budget constraints, is widely adopted by advertisers. A key challenge lies in designing algorithms for non-truthful mechanisms with ROI constraints. While prior work has addressed truthful auctions or non-truthful auctions with weaker benchmarks, this paper provides a significant improvement: We develop online bidding algorithms for repeated first-price auctions with ROI constraints, benchmarking against the optimal randomized strategy in hindsight. In the full feedback setting, where the maximum competing bid is observed, our algorithm achieves a near-optimal $\tilde O(\sqrt{T})$ regret bound, and in the bandit feedback setting (where the bidder only observes whether the bidder wins each auction), our algorithm attains $\tilde O(T^{3/4})$ regret bound.

Clock auctions are a natural class of simple auction mechanisms, with numerous desirable properties including obvious strategyproofness, credibility, and unconditional winner privacy. These properties make clock auctions an ideal solution for real-world allocation problems, such as radio spectrum (re)allocation. Accordingly, significant effort has been made in understanding the performance (and in particular, economic efficiency) of clock auctions. In this paper, we make progress along this direction on multiple fronts.

Bilateral trade is one of the most natural and important forms of economic interaction: A seller has a single, indivisible item for sale, and a buyer is potentially interested. The two parties typically have different, privately known valuations for the item, and ideally, they would like to trade if the buyer values the item more than the seller. The celebrated impossibility result by Myerson and Satterthwaite shows that any mechanism for this setting must violate at least one important desideratum. In this paper, we investigate a richer paradigm of bilateral trade, with many self-interested buyers and sellers on both sides of a single trade who cannot be excluded from the trade. We show that this allows for more positive results. In fact, we establish a dichotomy in the possibility of trading efficiently. If in expectation, the buyers value the item more, we can achieve efficiency in the limit. If this is not the case, then efficiency cannot be achieved in general. En route, we characterize trading mechanisms that encourage truth-telling, which may be of independent interest. We also evaluate our trading mechanisms experimentally, and the experiments align with our theoretical results.

We study the problem of online binary classification in settings where strategic agents can modify their observable features to receive a positive classification. We model the set of feasible manipulations by a directed graph over the feature space, and assume the learner only observes the manipulated features instead of the original ones. We introduce the Strategic Littlestone Dimension, a new combinatorial measure that captures the joint complexity of the hypothesis class and the manipulation graph. We demonstrate that it characterizes the instance-optimal mistake bounds for deterministic learning algorithms in the realizable setting. We also achieve improved regret in the agnostic setting by a refined agnostic-to-realizable reduction that accounts for the additional challenge of not observing agents' original features. Finally, we relax the assumption that the learner knows the manipulation graph, instead assuming their knowledge is captured by a family of graphs. We derive regret bounds in both the realizable setting where all agents manipulate according to the same graph within the graph family, and the agnostic setting where the manipulation graphs are chosen adversarially and not consistently modeled by a single graph in the family.

The recent increasing adoption of autobidding has inspired the growing interest in analyzing the performance of classic mechanism with value-maximizing autobidders both theoretically and empirically. It is known that optimal welfare can be obtained in first-price auctions if autobidders are restricted to uniform bid-scaling and the price of anarchy is $2$ when non-uniform bid-scaling strategies are allowed. In this paper, we provide a fine-grained price of anarchy analysis for non-uniform bid-scaling strategies in first-price auctions, demonstrating the reason why more powerful (individual) non-uniform bid-scaling strategies may lead to worse (aggregated) performance in social welfare. Our theoretical results match recent empirical findings that a higher level of non-uniform bid-scaling leads to lower welfare performance in first-price auctions.

We study the price of anarchy of first-price auctions in the autobidding world, where bidders can be either utility maximizers (i.e., traditional bidders) or value maximizers (i.e., autobidders). We show that with autobidders only, the price of anarchy of first-price auctions is $1/2$, and with both kinds of bidders, the price of anarchy degrades to about $0.457$ (the precise number is given by an optimization). These results complement the recent result by [Jin and Lu, 2022] showing that the price of anarchy of first-price auctions with traditional bidders is $1 - 1/e^2$. We further investigate a setting where the seller can utilize machine-learned advice to improve the efficiency of the auctions. There, we show that as the accuracy of the advice increases, the price of anarchy improves smoothly from about $0.457$ to $1$.

Quantitative Relative Judgment Aggregation (QRJA) is a new research topic in (computational) social choice. In the QRJA model, agents provide judgments on the relative quality of different candidates, and the goal is to aggregate these judgments across all agents. In this work, our main conceptual contribution is to explore the interplay between QRJA in a social choice context and its application to ranking prediction. We observe that in QRJA, judges do not have to be people with subjective opinions; for example, a race can be viewed as a ``judgment'' on the contestants' relative abilities. This allows us to aggregate results from multiple races to evaluate the contestants' true qualities. At a technical level, we introduce new aggregation rules for QRJA and study their structural and computational properties. We evaluate the proposed methods on data from various real races and show that QRJA-based methods offer effective and interpretable ranking predictions.

We study Bayesian persuasion under approximate best response, where the receiver may choose any action that is not too much suboptimal, given their posterior belief upon receiving the signal. We focus on the computational aspects of the problem, aiming to design algorithms that efficiently compute (almost) optimal strategies for the sender. Despite the absence of the revelation principle --- which has been one of the most powerful tools in Bayesian persuasion --- we design polynomial-time exact algorithms for the problem when either the state space or the action space is small, as well as a quasi-polynomial-time approximation scheme (QPTAS) for the general problem. On the negative side, we show there is no polynomial-time exact algorithm for the general problem unless $\mathsf{P} = \mathsf{NP}$. Our results build on several new algorithmic ideas, which might be useful in other principal-agent problems where robustness is desired.

We study the price of anarchy of the generalized second-price auction where bidders are value maximizers (i.e., autobidders). We show that in general the price of anarchy can be as bad as 0. For comparison, the price of anarchy of running VCG is 1/2 in the autobidding world. We further show a fined-grained price of anarchy with respect to the discount factors (i.e., the ratios of click probabilities between lower slots and the highest slot in each auction) in the generalized second-price auction, which highlights the qualitative relation between the smoothness of the discount factors and the efficiency of the generalized second-price auction.

We study equilibrium computation with extensive-form correlation in two-player turn-taking stochastic games. Our main results are two-fold: (1) We give an algorithm for computing a Stackelberg extensive-form correlated equilibrium (SEFCE), which runs in time polynomial in the size of the game, as well as the number of bits required to encode each input number. (2) We give an efficient algorithm for approximately computing an optimal extensive-form correlated equilibrium (EFCE) up to machine precision, i.e., the algorithm achieves approximation error ε in time polynomial in the size of the game, as well as log(1/ε).

We study autobidding ad auctions with user costs, where each bidder is value-maximizing subject to a return-over-investment (ROI) constraint, and the seller aims to maximize the social welfare taking into consideration the user’s cost of viewing an ad. We show that in the worst case, the approximation ratio of social welfare by running the vanilla VCG auctions with user costs could as bad as 0. To improve the performance of VCG, We propose a new variant of VCG based on properly chosen cost multipliers, and prove that there exist auction-dependent and bidder-dependent cost multipliers that guarantee approximation ratios of 1/2 and 1/4 respectively in terms of the social welfare.

The design of internet advertisement systems is both an auction design problem and an information retrieval (IR) problem. As an auction, the designer needs to take the participants incentives into account. As an information retrieval problem, it needs to identify the ad that it is the most relevant to a user out of an enormous set of ad candidates. Those aspects are combined by first having an IR system narrow down the initial set of ad candidates to a manageable size followed by an auction that ranks and prices those candidates.

We study the problem of learning influence adoption in networks. In this problem, a communicable entity (such as an infectious disease, a computer virus, or a social media meme) propagates through a network, and the goal is to learn the state of each individual node by sampling only a small number of nodes and observing/testing their states. We study this problem in heterogeneous networks, in which each individual node has a set of distinct features that determine how it is affected by the propagating entity. We give an efficient algorithm with nearly optimal sample complexity for two variants of this learning problem, corresponding to symptomatic and asymptomatic spread. In each case, the optimal sample complexity naturally generalizes the complexity of learning how nodes are affected in isolation, and the complexity of learning influence adoption in a homogeneous network.

We pose and study the problem of planning in Markov decision processes (MDPs), subject to participation constraints as studied in mechanism design. In this problem, a planner must work with a self-interested agent on a given MDP. Each action in the MDP provides an immediate reward to the planner and a (possibly different) reward to the agent. The agent has no control in choosing the actions, but has the option to end the entire process at any time. The goal of the planner is to find a policy that maximizes her cumulative reward, taking into consideration the agent's ability to terminate. We give a fully polynomial-time approximation scheme for this problem. En route, we present polynomial-time algorithms for computing (exact) optimal policies for important special cases of this problem, including when the time horizon is constant, or when the MDP exhibits a "definitive decisions" property. We illustrate our algorithms with two different game-theoretic applications: the problem of assigning rides in ride-sharing and the problem of designing screening policies. Our results imply efficient algorithms for computing (approximately) optimal policies in both applications.

Many scientific conferences employ a two-phase paper review process, where some papers are assigned additional reviewers after the initial reviews are submitted. Many conferences also design and run experiments on their paper review process, where some papers are assigned reviewers who provide reviews under an experimental condition. In this paper, we consider the question: how should reviewers be divided between phases or conditions in order to maximize total assignment similarity? We make several contributions towards answering this question. First, we prove that when the set of papers requiring additional review is unknown, a simplified variant of this problem is NP-hard. Second, we empirically show that across several datasets pertaining to real conference data, dividing reviewers between phases/conditions uniformly at random allows an assignment that is nearly as good as the oracle optimal assignment. This uniformly random choice is practical for both the two-phase and conference experiment design settings. Third, we provide explanations of this phenomenon by providing theoretical bounds on the suboptimality of this random strategy under certain natural conditions. From these easily-interpretable conditions, we provide actionable insights to conference program chairs about whether a random reviewer split is suitable for their conference.

We study posted price auctions and dynamic prior-independent mechanisms for (ROI-constrained) value maximizers. In contrast to classic (quasi-linear) utility maximizers, these agents aim to maximize their total value subject to a minimum ratio of value per unit of payment made. When personalized posted prices are allowed, posted price auctions for value maximizers can be reduced to posted price auctions for utility maximizers. However, for anonymous posted prices, the well-known $\frac 1 2$ approximation for utility maximizers is impossible for value maximizers and we provide a posted price mechanism with $\frac12(1 - 1/e)$ approximation. Moreover, we demonstrate how to apply our results to design prior-independent mechanisms in a dynamic environment; and to the best of our knowledge, this gives the first constant revenue approximation with multiple value maximizers. Finally, we provide an extension to combinatorial auctions with submodular / XOS agents.

We consider the problem of planning with participation constraints introduced in[24]. In this problem, a principal chooses actions in a Markov decision process, resulting in separate utilities for the principal and the agent. However, the agent can and will choose to end the process whenever his expected onward utility becomes negative. The principal seeks to compute and commit to a policy that maximizes her expected utility, under the constraint that the agent should always want to continue participating. We provide the first polynomial-time exact algorithm for this problem for finite-horizon settings, where previously only an additive ε-approximation algorithm was known. Our approach can also be extended to the (discounted) infinite-horizon case, for which we give an algorithm that runs in time polynomial in the size of the input and log(1/ε), and returns a policy that is optimal up to an additive error of ε.

Machine learning techniques can be useful in applications such as credit approval and college admission. However, to be classified more favorably in such contexts, an agent may decide to strategically withhold some of her features, such as bad test scores. This is a missing data problem with a twist: which data is missing depends on the chosen classifier, because the specific classifier is what may create the incentive to withhold certain feature values. We address the problem of training classifiers that are robust to this behavior. We design three classification methods: MINCUT, Hill-Climbing (HC) and Incentive-Compatible Logistic Regression (IC-LR). We show that MINCUT is optimal when the true distribution of data is fully known. However, it can produce complex decision boundaries, and hence be prone to overfitting in some cases. Based on a characterization of truthful classifiers (i.e., those that give no incentive to strategically hide features), we devise a simpler alternative called HC which consists of a hierarchical ensemble of out-of-the-box classifiers, trained using a specialized hill-climbing procedure which we show to be convergent. For several reasons, MINCUT and HC are not effective in utilizing a large number of complementarily informative features. To this end, we present IC-LR, a modification of Logistic Regression that removes the incentive to strategically drop features. We also show that our algorithms perform well in experiments on real-world data sets, and present insights into their relative performance in different settings.

We study the problem of automated mechanism design with partial verification, where each type can (mis)report only a restricted set of types (rather than any other type), induced by the principal's limited verification power. We prove hardness results when the revelation principle does not necessarily hold, as well as when types have even minimally different preferences. In light of these hardness results, we focus on truthful mechanisms in the setting where all types share the same preference over outcomes, which is motivated by applications in, e.g., strategic classification. We present a number of algorithmic and structural results, including an efficient algorithm for finding optimal deterministic truthful mechanisms, which also implies a faster algorithm for finding optimal randomized truthful mechanisms via a characterization based on the notion of convexity. We then consider a more general setting, where the principal's cost is a function of the combination of outcomes assigned to each type. In particular, we focus on the case where the cost function is submodular, and give generalizations of essentially all our results in the classical setting where the cost function is additive. Our results provide a relatively complete picture for automated mechanisms design with partial verification.

We study test-based binary classification, where a principal either accepts or rejects agents based on the outcomes they get in a set of tests. The principal commits to a policy, which consists of all sets of outcomes that lead to acceptance. Each agent is modeled by a distribution over the space of possible outcomes. When an agent takes a test, he pays a cost and receives an independent sample from his distribution as the outcome. Agents can always choose between taking another test and stopping. They maximize their expected utility, which is the value of acceptance if the principal's policy accepts the set of outcomes they have and 0 otherwise, minus the total cost of tests taken. We focus on the case where agents can be either "good" or "bad" (corresponding to their distribution over test outcomes), and the principal's goal is to accept good agents and reject bad ones. We show, roughly speaking, that as long as the good and bad agents have different distributions (which can be arbitrarily close to each other), the principal can always achieve perfect accuracy, meaning good agents are accepted with probability 1, and bad ones are rejected with probability 1. Moreover, there is a policy achieving perfect accuracy under which the maximum number of tests any agent needs to take is constant — in sharp contrast to the case where the principal directly observes samples from agents' distributions. The key technique is to choose the policy so that agents self-select into taking tests.

We study PAC learning in the presence of strategic manipulation, where data points may modify their features in certain predefined ways in order to receive a better outcome. We show that the vanilla ERM principle fails to achieve any nontrivial guarantee in this context. Instead, we propose an incentive-aware version of the ERM principle which has asymptotically optimal sample complexity. We then focus our attention on incentive-compatible classifiers, which provably prevent any kind of strategic manipulation. We give a sample complexity bound that is, curiously, independent of the hypothesis class, for the ERM principle restricted to incentive-compatible classifiers. This suggests that incentive compatibility alone can act as an effective means of regularization. We further show that it is without loss of generality to consider only incentive-compatible classifiers when opportunities for strategic manipulation satisfy a transitivity condition. As a consequence, in such cases, our hypothesis-class-independent sample complexity bound applies even without incentive compatibility. Our results set the foundations of incentive-aware PAC learning.

We study prior-independent dynamic auction design with production costs for a value-maximizing buyer, a paradigm that is becoming prevalent recently following the development of automatic bidding algorithms in advertising platforms. In contrast to a utility-maximizing buyer, who maximizes the difference between her total value and total payment, a value-maximizing buyer aims to maximize her total value subject to a return on investment (ROI) constraint. Our main result is a dynamic mechanism with regret $\tilde{O}(T^{2/3})$, where $T$ is the time horizon, against the first-best benchmark, i.e., the maximum amount of revenue the seller can extract assuming all values of the buyer are publicly known.

We study Bayesian automated mechanism design in unstructured dynamic environments, where a principal repeatedly interacts with an agent, and takes actions based on the strategic agent's report of the current state of the world. Both the principal and the agent can have arbitrary and potentially different valuations for the actions taken, possibly also depending on the actual state of the world. Moreover, at any time, the state of the world may evolve arbitrarily depending on the action taken by the principal. The goal is to compute an optimal mechanism which maximizes the principal's utility in the face of the self-interested strategic agent.We give an efficient algorithm for computing optimal mechanisms, with or without payments, under different individual-rationality constraints, when the time horizon is constant. Our algorithm is based on a sophisticated linear program formulation, which can be customized in various ways to accommodate richer constraints. For environments with large time horizons, we show that the principal's optimal utility is hard to approximate within a certain constant factor, complementing our algorithmic result. These results paint a relatively complete picture for automated dynamic mechanism design in unstructured environments. We further consider a special case of the problem where the agent is myopic, and give a refined efficient algorithm whose time complexity scales linearly in the time horizon. In the full version of the paper, we show that memoryless mechanisms, which are without loss of generality optimal in Markov decision processes without strategic behavior, do not provide a good solution for our problem, in terms of both optimality and computational tractability. Moreover, we present experimental results where our algorithms are applied to synthetic dynamic environments with different characteristics, which not only serve as a proof of concept for our algorithms, but also exhibit intriguing phenomena in dynamic mechanism design.

We study combinatorial auctions in online environments with the goal of maximizing social welfare. In this problem, new items become available on each day and must be sold before their respective expiration dates. We design online auctions for the widely studied classes of submodular and XOS valuations, and show the following results: – For submodular valuations, we give an O(log m)-competitive mechanism for adversarial valuations and an O(1)-competitive mechanism for Bayesian valuations, where m is the total number of items. Both these mechanisms are computationally efficient and universally truthful for myopic agents, i.e., agents with no knowledge of the future. – For XOS valuations, we show that there is no online mechanism that can achieve a competitive ratio of o ((m/ log m)1/3) even in a Bayesian setting. Our lower bound holds even if we do not require truthfulness and/or computational efficiency of the mechanism. This establishes a sharp separation between XOS valuations and its subclass of submodular valuations for online combinatorial auctions. In contrast, no such separation exists for offline auctions, where the best bounds for both submodular and XOS valuations are O((log log m)3) for adversarial settings (Assadi and Singla, FOCS 2019) and O(1) for Bayesian settings (Dütting et al., FOCS 2017). In contrast to the above, if items do not expire and only need to be sold before the market closes, then we give a reduction from offline to online mechanisms that preserves the competitive ratio for all subadditive valuations (that includes XOS and submodular valuations), thereby achieving the same bounds as the respective best offline mechanisms.

We give a framework for designing prophet inequalities for combinatorial welfare maximization. Instantiated with different parameters, our framework implies (1) an O(log m / log log m)-competitive prophet inequality for subadditive agents, improving over the O(log m) upper bound via item pricing, (2) an O(D log m / log log m)-competitive prophet inequality for D-approximately subadditive agents, where D ∈ {1, … , m-1} measures the maximum number of items that complement each other, and (3) as a byproduct, an O(1)-competitive prophet inequality for submodular or fractionally subadditive (a.k.a. XOS) agents, matching the optimal ratio asymptotically. Our framework is computationally efficient given sample access to the prior and demand queries.

We study the problem of learning opinions in social networks. The learner observes the states of some sample nodes from a social network, and tries to infer the states of other nodes, based on the structure of the network. We show that sample-efficient learning is impossible when the network exhibits strong noise, and give a polynomial-time algorithm for the problem with nearly optimal sample complexity when the network is sufficiently stable.

We give a row sampling algorithm for the quantile loss function with sample complexity nearly linear in the dimensionality of the data, improving upon the previous best algorithm whose sampling complexity has at least cubic dependence on the dimensionality. Based upon our row sampling algorithm, we give the fastest known algorithm for quantile regression and a graph sparsification algorithm for balanced directed graphs. Our main technical contribution is to show that Lewis weights sampling, which has been used in row sampling algorithms for $\ell_p$ norms, can also be applied in row sampling algorithms for a variety of loss functions. We complement our theoretical results by experiments to demonstrate the practicality of our approach.

We study problems where a learner aims to learn the valuations of an agent by observing which goods he buys under varying price vectors. More specifically, we consider the case of a $k$-demand agent, whose valuation over the goods is additive when receiving up to $k$ goods, but who has no interest in receiving more than $k$ goods. We settle the query complexity for the active-learning (preference elicitation) version, where the learner chooses the prices to post, by giving a \emph{biased binary search} algorithm, generalizing the classical binary search procedure. We complement our query complexity upper bounds by lower bounds that match up to lower-order terms. We also study the passive-learning version in which the learner does not control the prices, and instead they are sampled from some distribution. We show that in the PAC model for passive learning, any \emph{empirical risk minimizer} has a sample complexity that is optimal up to a factor of $\widetilde{O}(k)$.

We consider three important challenges in conference peer review: (i) reviewers maliciously attempting to get assigned to certain papers to provide positive reviews, possibly as part of quid-pro-quo arrangements with the authors; (ii) "torpedo reviewing," where reviewers deliberately attempt to get assigned to certain papers that they dislike in order to reject them; (iii) reviewer de-anonymization on release of the similarities and the reviewer-assignment code. On the conceptual front, we identify connections between these three problems and present a framework that brings all these challenges under a common umbrella. We then present a (randomized) algorithm for reviewer assignment that can optimally solve the reviewer-assignment problem under any given constraints on the probability of assignment for any reviewer-paper pair. We further consider the problem of restricting the joint probability that certain suspect pairs of reviewers are assigned to certain papers, and show that this problem is NP-hard for arbitrary constraints on these joint probabilities but efficiently solvable for a practical special case. Finally, we experimentally evaluate our algorithms on datasets from past conferences, where we observe that they can limit the chance that any malicious reviewer gets assigned to their desired paper to 50% while producing assignments with over 90% of the total optimal similarity.

We consider a combinatorial variant of the classical ski rental problem --- which we call combinatorial ski rental --- where multiple resources are available to purchase and to rent, and are demanded online. Moreover, the costs of purchasing and renting are potentially combinatorial. The dual problem of combinatorial ski rental, which we call combinatorial online bipartite matching, generalizes the classical online bipartite matching problem into a form where constraints, induced by both offline and online vertices, can be combinatorial. We give a 2-competitive (resp. e / (e - 1)-competitive) deterministic (resp. randomized) algorithm for combinatorial ski rental, and an e / (e - 1)-competitive algorithm for combinatorial online bipartite matching. All these ratios are optimal given simple lower bounds inherited from the respective well-studied special cases. We also prove information-theoretic impossibility of constant-factor algorithms when any part of our assumptions is considerably relaxed.

We study PMAC-learning of real-valued set functions with limited complementarity. We prove, to our knowledge, the first nontrivial learnability result for set functions exhibiting complementarity, generalizing Balcan and Harvey’s result for submodular functions. We prove a nearly matching information theoretical lower bound on the number of samples required, complementing our learnability result. We conduct numerical simulations to show that our algorithm is likely to perform well in practice.

Specifying the objective function that an AI system should pursue can be challenging. Especially when the decisions to be made by the system have a moral component, input from multiple stakeholders is often required. We consider approaches that query them about their judgments in individual examples, and then aggregate these judgments into a general policy. We propose a formal learning-theoretic framework for this setting. We then give general results on how to translate classical results from PAC learning into results in our framework. Subsequently, we show that in some settings, better results can be obtained by working directly in our framework. Finally, we discuss how our model can be extended in a variety of ways for future research.

In standard classification problems, the assumption is that the entity making the decision (the principal) has access to all the samples. However, in many contexts, she either does not have direct access to the samples, or can inspect only a limited set of samples and does not know which are the most relevant ones. In such cases, she must rely on another party (the agent) to either provide the samples or point out the most relevant ones. If the agent has a different objective, then the principal cannot trust the submitted samples to be representative. She must set a policy for how she makes decisions, keeping in mind the agent’s incentives. In this paper, we introduce a theoretical framework for this problem and provide key structural and computational results.

We introduce two new "degree of complementarity" measures: supermodular width and superadditive width. Both are formulated based on natural witnesses of complementarity. We show that both measures are robust by proving that they, respectively, characterize the gap of monotone set functions from being submodular and subadditive. Thus, they define two new hierarchies over monotone set functions, which we will refer to as Supermodular Width (SMW) hierarchy and Superadditive Width (SAW) hierarchy, with foundations - i.e. level 0 of the hierarchies - resting exactly on submodular and subadditive functions, respectively. We present a comprehensive comparative analysis of the SMW hierarchy and the Supermodular Degree (SD) hierarchy, defined by Feige and Izsak. We prove that the SMW hierarchy is strictly more expressive than the SD hierarchy: Every monotone set function of supermodular degree d has supermodular width at most d, and there exists a supermodular-width-1 function over a ground set of m elements whose supermodular degree is m-1. We show that previous results regarding approximation guarantees for welfare and constrained maximization as well as regarding the Price of Anarchy (PoA) of simple auctions can be extended without any loss from the supermodular degree to the supermodular width. We also establish almost matching information-theoretical lower bounds for these two well-studied fundamental maximization problems over set functions. The combination of these approximation and hardness results illustrate that the SMW hierarchy provides not only a natural notion of complementarity, but also an accurate characterization of "near submodularity" needed for maximization approximation. While SD and SMW hierarchies support nontrivial bounds on the PoA of simple auctions, we show that our SAW hierarchy seems to capture more intrinsic properties needed to realize the efficiency of simple auctions. So far, the SAW hierarchy provides the best dependency for the PoA of Single-bid Auction, and is nearly as competitive as the Maximum over Positive Hypergraphs (MPH) hierarchy for Simultaneous Item First Price Auction (SIA). We also provide almost tight lower bounds for the PoA of both auctions with respect to the SAW hierarchy.

Q-learning with function approximation is one of the most popular methods in reinforcement learning. Though the idea of using function approximation was proposed at least 60 years ago, even in the simplest setup, i.e, approximating Q-functions with linear functions, it is still an open problem how to design a provably efficient algorithm that learns a near-optimal policy. The key challenges are how to efficiently explore the state space and how to decide when to stop exploring in conjunction with the function approximation scheme. The current paper presents a provably efficient algorithm for Q-learning with linear function approximation. Under certain regularity assumptions, our algorithm, Difference Maximization Q-learning, combined with linear function approximation, returns a near-optimal policy using polynomial number of trajectories. Our algorithm introduces a new notion, the Distribution Shift Error Checking (DSEC) oracle. This oracle tests whether there exists a function in the function class that predicts well on a distribution $\mathcal{D}_1$, but predicts poorly on another distribution $\mathcal{D}_2$, where $\mathcal{D}_1$ and $\mathcal{D}_2$ are distributions over states induced by two different exploration policies. For the linear function class, this oracle is equivalent to solving a top eigenvalue problem. We believe our algorithmic insights, especially the DSEC oracle, are also useful in designing and analyzing reinforcement learning algorithms with general function approximation.

Often, a principal must make a decision based on data provided by an agent. Moreover, typically, that agent has an interest in the decision that is not perfectly aligned with that of the principal. Thus, the agent may have an incentive to select from or modify the samples he obtains before sending them to the principal. In other settings, the principal may not even be able to observe samples directly; instead, she must rely on signals that the agent is able to send based on the samples that he obtains, and he will choose these signals strategically. In this paper, we give necessary and sufficient conditions for when the principal can distinguish between agents of good'' andbad'' types, when the type affects the distribution of samples that the agent has access to. We also study the computational complexity of checking these conditions. Finally, we study how many samples are needed.

We revisit the well-studied problem of constructing strategyproof approximation mechanisms for facility location games, but offer a fundamentally new perspective by considering risk averse designers. Specifically, we are interested in the tradeoff between a randomized strategyproof mechanism's approximation ratio, and its variance (which has long served as a proxy for risk). When there is just one facility, we observe that the social cost objective is trivial, and derive the optimal tradeoff with respect to the maximum cost objective. When there are multiple facilities, the main challenge is the social cost objective, and we establish a surprising impossibility result: under mild assumptions, no smooth approximation-variance tradeoff exists. We also discuss the implications of our work for computational mechanism design at large.