arXiv subjects

Moshe Tennenholtz

Publications and source records attributed to Moshe Tennenholtz.

At least 91 records · Page 5Linked to original sources

arXiv2018

Recommendation Systems and Self Motivated Users

Modern recommendation systems rely on the wisdom of the crowd to learn the optimal course of action. This induces an inherent mis-alignment of incentives between the system's objective to learn (explore) and the individual users' objective to take the contemporaneous optimal action (exploit). The design of such systems must account for this and also for additional information available to the users. A prominent, yet simple, example is when agents arrive sequentially and each agent observes the action and reward of his predecessor. We provide an incentive compatible and asymptotically optimal mechanism for that setting. The complexity of the mechanism suggests that the design of such systems for general settings is a challenging task.

cs.GT↗

arXiv2018

Ranking Robustness Under Adversarial Document Manipulations

For many queries in the Web retrieval setting there is an on-going ranking competition: authors manipulate their documents so as to promote them in rankings. Such competitions can have unwarranted effects not only in terms of retrieval effectiveness, but also in terms of ranking robustness. A case in point, rankings can (rapidly) change due to small indiscernible perturbations of documents. While there has been a recent growing interest in analyzing the robustness of classifiers to adversarial manipulations, there has not yet been a study of the robustness of relevance-ranking functions. We address this challenge by formally analyzing different definitions and aspects of the robustness of learning-to-rank-based ranking functions. For example, we formally show that increased regularization of linear ranking functions increases ranking robustness. This finding leads us to conjecture that decreased variance of any ranking function results in increased robustness. We propose several measures for quantifying ranking robustness and use them to analyze ranking competitions between documents' authors. The empirical findings support our formal analysis and conjecture for both RankSVM and LambdaMART.

cs.IR↗

arXiv2018

Segmentation, Incentives and Privacy

Data driven segmentation is the powerhouse behind the success of online advertising. Various underlying challenges for successful segmentation have been studied by the academic community, with one notable exception - consumers incentives have been typically ignored. This lacuna is troubling as consumers have much control over the data being collected. Missing or manipulated data could lead to inferior segmentation. The current work proposes a model of prior-free segmentation, inspired by models of facility location, and to the best of our knowledge provides the first segmentation mechanism that addresses incentive compatibility, efficient market segmentation and privacy in the absence of a common prior.

cs.GT↗

arXiv2018

Game of Coins

We formalize the current practice of strategic mining in multi-cryptocurrency markets as a game, and prove that any better-response learning in such games converges to equilibrium. We then offer a reward design scheme that moves the system configuration from any initial equilibrium to a desired one for any better-response learning of the miners. Our work introduces the first multi-coin strategic attack for adaptive and learning miners, as well as the study of reward design in a multi-agent system of learning agents.

cs.GT↗

arXiv2018

Incentive-Compatible Diffusion

Our work bridges the literature on incentive-compatible mechanism design and the literature on diffusion algorithms. We introduce the study of finding an incentive-compatible (strategy-proof) mechanism for selecting an influential vertex in a directed graph (e.g. Twitter's network). The goal is to devise a mechanism with a bounded ratio between the maximal influence and the influence of the selected user, and in which no user can improve its probability of being selected by following or unfollowing other users. We introduce the `Two Path' mechanism which is based on the idea of selecting the vertex that is the first intersection of two independent random walks in the network. The Two Path mechanism is incentive compatible on directed acyclic graphs (DAGs), and has a finite approximation ratio on natural subfamilies of DAGs. Simulations indicate that this mechanism is suitable for practical uses.

cs.GT↗

arXiv2018

Characterizing Efficient Referrals in Social Networks

Users of social networks often focus on specific areas of that network, leading to the well-known "filter bubble" effect. Connecting people to a new area of the network in a way that will cause them to become active in that area could help alleviate this effect and improve social welfare. Here we present preliminary analysis of network referrals, that is, attempts by users to connect peers to other areas of the network. We classify these referrals by their efficiency, i.e., the likelihood that a referral will result in a user becoming active in the new area of the network. We show that by using features describing past experience of the referring author and the content of their messages we are able to predict whether referral will be effective, reaching an AUC of 0.87 for those users most experienced in writing efficient referrals. Our results represent a first step towards algorithmically constructing efficient referrals with the goal of mitigating the "filter bubble" effect pervasive in on line social networks.

cs.SI↗

arXiv2017

Shapley Facility Location Games

Facility location games have been a topic of major interest in economics, operations research and computer science, starting from the seminal work by Hotelling. Spatial facility location models have successfully predicted the outcome of competition in a variety of scenarios. In a typical facility location game, users/customers/voters are mapped to a metric space representing their preferences, and each player picks a point (facility) in that space. In most facility location games considered in the literature, users are assumed to act deterministically: given the facilities chosen by the players, users are attracted to their nearest facility. This paper introduces facility location games with probabilistic attraction, dubbed Shapley facility location games, due to a surprising connection to the Shapley value. The specific attraction function we adopt in this model is aligned with the recent findings of the behavioral economics literature on choice prediction. Given this model, our first main result is that Shapley facility location games are potential games; hence, they possess pure Nash equilibrium. Moreover, the latter is true for any compact user space, any user distribution over that space, and any number of players. Note that this is in sharp contrast to Hotelling facility location games. In our second main result we show that under the assumption that players can compute an approximate best response, approximate equilibrium profiles can be learned efficiently by the players via dynamics. Our third main result is a bound on the Price of Anarchy of this class of games, as well as showing the bound is tight. Ultimately, we show that player payoffs coincide with their Shapley value in a coalition game, where coalition gains are the social welfare of the users.

cs.GT↗

arXiv2017

Group Recommendations: Axioms, Impossibilities, and Random Walks

We introduce an axiomatic approach to group recommendations, in line of previous work on the axiomatic treatment of trust-based recommendation systems, ranking systems, and other foundational work on the axiomatic approach to internet mechanisms in social choice settings. In group recommendations we wish to recommend to a group of agents, consisting of both opinionated and undecided members, a joint choice that would be acceptable to them. Such a system has many applications, such as choosing a movie or a restaurant to go to with a group of friends, recommending games for online game players, & other communal activities. Our method utilizes a given social graph to extract information on the undecided, relying on the agents influencing them. We first show that a set of fairly natural desired requirements (a.k.a axioms) leads to an impossibility, rendering mutual satisfaction of them unreachable. However, we also show a modified set of axioms that fully axiomatize a group variant of the random-walk recommendation system, expanding a previous result from the individual recommendation case.

cs.SI↗

arXiv2017

ERA: A Framework for Economic Resource Allocation for the Cloud

Cloud computing has reached significant maturity from a systems perspective, but currently deployed solutions rely on rather basic economics mechanisms that yield suboptimal allocation of the costly hardware resources. In this paper we present Economic Resource Allocation (ERA), a complete framework for scheduling and pricing cloud resources, aimed at increasing the efficiency of cloud resources usage by allocating resources according to economic principles. The ERA architecture carefully abstracts the underlying cloud infrastructure, enabling the development of scheduling and pricing algorithms independently of the concrete lower-level cloud infrastructure and independently of its concerns. Specifically, ERA is designed as a flexible layer that can sit on top of any cloud system and interfaces with both the cloud resource manager and with the users who reserve resources to run their jobs. The jobs are scheduled based on prices that are dynamically calculated according to the predicted demand. Additionally, ERA provides a key internal API to pluggable algorithmic modules that include scheduling, pricing and demand prediction. We provide a proof-of-concept software and demonstrate the effectiveness of the architecture by testing ERA over both public and private cloud systems -- Azure Batch of Microsoft and Hadoop/YARN. A broader intent of our work is to foster collaborations between economics and system communities. To that end, we have developed a simulation platform via which economics and system experts can test their algorithmic implementations.

cs.GT↗

arXiv2016

An Axiomatic Approach to Routing

Information delivery in a network of agents is a key issue for large, complex systems that need to do so in a predictable, efficient manner. The delivery of information in such multi-agent systems is typically implemented through routing protocols that determine how information flows through the network. Different routing protocols exist each with its own benefits, but it is generally unclear which properties can be successfully combined within a given algorithm. We approach this problem from the axiomatic point of view, i.e., we try to establish what are the properties we would seek to see in such a system, and examine the different properties which uniquely define common routing algorithms used today. We examine several desirable properties, such as robustness, which ensures adding nodes and edges does not change the routing in a radical, unpredictable ways; and properties that depend on the operating environment, such as an "economic model", where nodes choose their paths based on the cost they are charged to pass information to the next node. We proceed to fully characterize minimal spanning tree, shortest path, and weakest link routing algorithms, showing a tight set of axioms for each.

cs.GT↗

arXiv2016

A Hydraulic Approach to Equilibria of Resource Selection Games

Drawing intuition from a (physical) hydraulic system, we present a novel framework, constructively showing the existence of a strong Nash equilibrium in resource selection games (i.e., asymmetric singleton congestion games) with nonatomic players, the coincidence of strong equilibria and Nash equilibria in such games, and the uniqueness of the cost of each given resource across all Nash equilibria. Our proofs allow for explicit calculation of Nash equilibrium and for explicit and direct calculation of the resulting (unique) costs of resources, and do not hinge on any fixed-point theorem, on the Minimax theorem or any equivalent result, on linear programming, or on the existence of a potential (though our analysis does provide powerful insights into the potential, via a natural concrete physical interpretation). A generalization of resource selection games, called resource selection games with I.D.-dependent weighting, is defined, and the results are extended to this family, showing the existence of strong equilibria, and showing that while resource costs are no longer unique across Nash equilibria in games of this family, they are nonetheless unique across all strong Nash equilibria, drawing a novel fundamental connection between group deviation and I.D.-congestion. A natural application of the resulting machinery to a large class of constraint-satisfaction problems is also described.

cs.GT↗

arXiv2016

Dynamics of Evolving Social Groups

Exclusive social groups are ones in which the group members decide whether or not to admit a candidate to the group. Examples of exclusive social groups include academic departments and fraternal organizations. In the present paper we introduce an analytic framework for studying the dynamics of exclusive social groups. In our model, every group member is characterized by his opinion, which is represented as a point on the real line. The group evolves in discrete time steps through a voting process carried out by the group's members. Due to homophily, each member votes for the candidate who is more similar to him (i.e., closer to him on the line). An admission rule is then applied to determine which candidate, if any, is admitted. We consider several natural admission rules including majority and consensus. We ask: how do different admission rules affect the composition of the group in the long term? We study both growing groups (where new members join old ones) and fixed-size groups (where new members replace those who quit). Our analysis reveals intriguing phenomena and phase transitions, some of which are quite counterintuitive.

cs.GT↗

arXiv2016

A Reinforcement Learning System to Encourage Physical Activity in Diabetes Patients

Regular physical activity is known to be beneficial to people suffering from diabetes type 2. Nevertheless, most such people are sedentary. Smartphones create new possibilities for helping people to adhere to their physical activity goals, through continuous monitoring and communication, coupled with personalized feedback. We provided 27 sedentary diabetes type 2 patients with a smartphone-based pedometer and a personal plan for physical activity. Patients were sent SMS messages to encourage physical activity between once a day and once per week. Messages were personalized through a Reinforcement Learning (RL) algorithm which optimized messages to improve each participant's compliance with the activity regimen. The RL algorithm was compared to a static policy for sending messages and to weekly reminders. Our results show that participants who received messages generated by the RL algorithm increased the amount of activity and pace of walking, while the control group patients did not. Patients assigned to the RL algorithm group experienced a superior reduction in blood glucose levels (HbA1c) compared to control policies, and longer participation caused greater reductions in blood glucose levels. The learning algorithm improved gradually in predicting which messages would lead participants to exercise. Our results suggest that a mobile phone application coupled with a learning algorithm can improve adherence to exercise in diabetic patients. As a learning algorithm is automated, and delivers personalized messages, it could be used in large populations of diabetic patients to improve health and glycemic control. Our results can be expanded to other areas where computer-led health coaching of humans may have a positive impact.

cs.CY↗

arXiv2015

Economic Recommendation Systems

In the on-line Explore and Exploit literature, central to Machine Learning, a central planner is faced with a set of alternatives, each yielding some unknown reward. The planner's goal is to learn the optimal alternative as soon as possible, via experimentation. A typical assumption in this model is that the planner has full control over the experiment design and implementation. When experiments are implemented by a society of self-motivated agents the planner can only recommend experimentation but has no power to enforce it. Kremer et al (JPE, 2014) introduce the first study of explore and exploit schemes that account for agents' incentives. In their model it is implicitly assumed that agents do not see nor communicate with each other. Their main result is a characterization of an optimal explore and exploit scheme. In this work we extend Kremer et al (JPE, 2014) by adding a layer of a social network according to which agents can observe each other. It turns out that when observability is factored in the scheme proposed by Kremer et al (JPE, 2014) is no longer incentive compatible. In our main result we provide a tight bound on how many other agents can each agent observe and still have an incentive-compatible algorithm and asymptotically optimal outcome. More technically, for a setting with N agents where the number of nodes with degree greater than N^alpha is bounded by N^beta and 2*alpha+beta < 1 we construct incentive-compatible asymptotically optimal mechanism. The bound 2*alpha+beta < 1 is shown to be tight.

cs.GT↗

arXiv2015

Distributed Signaling Games

A recurring theme in recent computer science literature is that proper design of signaling schemes is a crucial aspect of effective mechanisms aiming to optimize social welfare or revenue. One of the research endeavors of this line of work is understanding the algorithmic and computational complexity of designing efficient signaling schemes. In reality, however, information is typically not held by a central authority, but is distributed among multiple sources (third-party "mediators"), a fact that dramatically changes the strategic and combinatorial nature of the signaling problem, making it a game between information providers, as opposed to a traditional mechanism design problem. In this paper we introduce {\em distributed signaling games}, while using display advertising as a canonical example for introducing this foundational framework. A distributed signaling game may be a pure coordination game (i.e., a distributed optimization task), or a non-cooperative game. In the context of pure coordination games, we show a wide gap between the computational complexity of the centralized and distributed signaling problems. On the other hand, we show that if the information structure of each mediator is assumed to be "local", then there is an efficient algorithm that finds a near-optimal ($5$-approximation) distributed signaling scheme. In the context of non-cooperative games, the outcome generated by the mediators' signals may have different value to each (due to the auctioneer's desire to align the incentives of the mediators with his own by relative compensations). We design a mechanism for this problem via a novel application of Shapley's value, and show that it possesses some interesting properties, in particular, it always admits a pure Nash equilibrium, and it never decreases the revenue of the auctioneer.

cs.GT↗

arXiv2015

A Mirage of Market Allocation

Can noncooperative behaviour of merchants lead to a market split that prima facie seems anticompetitive? We introduce a model in which service providers, with ISPs being the main example, aim at optimizing the number of customers using their services, while customers aim at choosing service providers with low customer load (high bandwidth per subscriber, for ISPs). Each service provider chooses between a variety of levels of service (latencies, for ISPs), and as long as it does not lose customers, aims at minimizing its level of service; the minimum level of service required to satisfy a customer varies across customers. We consider a two-stage competition: in the first stage, service providers select their levels of service; in the second stage, customers choose between service providers. In the two-stage game, we show that the competition among service providers possesses a unique Nash equilibrium, which is moreover super-strong; we also show that sequential better-response dynamics of service providers reach this equilibrium, with best-response dynamics doing so surprisingly fast. If service providers choose their levels of service according to this equilibrium, then the unique Nash equilibrium among customers in the second phase is a split of the market between the service providers, based on the customers' minimum acceptable quality of service; moreover, each service provider's chosen level of service is the lowest acceptable by the entirety of its market slice, seemingly making no attempt to attract other customers. Our results show that this prima facie market allocation (collusive split of the market) arises as the unique and highly robust outcome of noncooperative, even myopic, service-provider behaviour. These results are applicable to a wide variety of scenarios, from explaining phenomena observable in food markets, to shedding a surprising light on aspects of location theory.

cs.GT↗

arXiv2015

Mechanism Design with Strategic Mediators

We consider the problem of designing mechanisms that interact with strategic agents through strategic intermediaries (or mediators), and investigate the cost to society due to the mediators' strategic behavior. Selfish agents with private information are each associated with exactly one strategic mediator, and can interact with the mechanism exclusively through that mediator. Each mediator aims to optimize the combined utility of his agents, while the mechanism aims to optimize the combined utility of all agents. We focus on the problem of facility location on a metric induced by a publicly known tree. With non-strategic mediators, there is a dominant strategy mechanism that is optimal. We show that when both agents and mediators act strategically, there is no dominant strategy mechanism that achieves any approximation. We, thus, slightly relax the incentive constraints, and define the notion of a two-sided incentive compatible mechanism. We show that the $3$-competitive deterministic mechanism suggested by Procaccia and Tennenholtz (2013) and Dekel et al. (2010) for lines extends naturally to trees, and is still $3$-competitive as well as two-sided incentive compatible. This is essentially the best possible. We then show that by allowing randomization one can construct a $2$-competitive randomized mechanism that is two-sided incentive compatible, and this is also essentially tight. This result also closes a gap left in the work of Procaccia and Tennenholtz (2013) and Lu et al. (2009) for the simpler problem of designing strategy-proof mechanisms for weighted agents with no mediators on a line, while extending to the more general model of trees. We also investigate a further generalization of the above setting where there are multiple levels of mediators.

cs.GT↗

arXiv2014

Chasing Ghosts: Competing with Stateful Policies

We consider sequential decision making in a setting where regret is measured with respect to a set of stateful reference policies, and feedback is limited to observing the rewards of the actions performed (the so called "bandit" setting). If either the reference policies are stateless rather than stateful, or the feedback includes the rewards of all actions (the so called "expert" setting), previous work shows that the optimal regret grows like $Θ(\sqrt{T})$ in terms of the number of decision rounds $T$. The difficulty in our setting is that the decision maker unavoidably loses track of the internal states of the reference policies, and thus cannot reliably attribute rewards observed in a certain round to any of the reference policies. In fact, in this setting it is impossible for the algorithm to estimate which policy gives the highest (or even approximately highest) total reward. Nevertheless, we design an algorithm that achieves expected regret that is sublinear in $T$, of the form $O( T/\log^{1/4}{T})$. Our algorithm is based on a certain local repetition lemma that may be of independent interest. We also show that no algorithm can guarantee expected regret better than $O( T/\log^{3/2} T)$.

cs.LG↗