Models, code, and papers for "Mingyan Liu":

Making judicious channel access and transmission scheduling decisions is essential for improving performance as well as energy and spectral efficiency in multichannel wireless systems. This problem has been a subject of extensive study in the past decade, and the resulting dynamic and opportunistic channel access schemes can bring potentially significant improvement over traditional schemes. However, a common and severe limitation of these dynamic schemes is that they almost always require some form of a priori knowledge of the channel statistics. A natural remedy is a learning framework, which has also been extensively studied in the same context, but a typical learning algorithm in this literature seeks only the best static policy, with performance measured by weak regret, rather than learning a good dynamic channel access policy. There is thus a clear disconnect between what an optimal channel access policy can achieve with known channel statistics that actively exploits temporal, spatial and spectral diversity, and what a typical existing learning algorithm aims for, which is the static use of a single channel devoid of diversity gain. In this paper we bridge this gap by designing learning algorithms that track known optimal or sub-optimal dynamic channel access and transmission scheduling policies, thereby yielding performance measured by a form of strong regret, the accumulated difference between the reward returned by an optimal solution when a priori information is available and that by our online algorithm. We do so in the context of two specific algorithms that appeared in [1] and [2], respectively, the former for a multiuser single-channel setting and the latter for a single-user multichannel setting. In both cases we show that our algorithms achieve sub-linear regret uniform in time and outperforms the standard weak-regret learning algorithms.

We analyze the following group learning problem in the context of opinion diffusion: Consider a network with $M$ users, each facing $N$ options. In a discrete time setting, at each time step, each user chooses $K$ out of the $N$ options, and receive randomly generated rewards, whose statistics depend on the options chosen as well as the user itself, and are unknown to the users. Each user aims to maximize their expected total rewards over a certain time horizon through an online learning process, i.e., a sequence of exploration (sampling the return of each option) and exploitation (selecting empirically good options) steps. Within this context we consider two group learning scenarios, (1) users with uniform preferences and (2) users with diverse preferences, and examine how a user should construct its learning process to best extract information from other's decisions and experiences so as to maximize its own reward. Performance is measured in {\em weak regret}, the difference between the user's total reward and the reward from a user-specific best single-action policy (i.e., always selecting the set of options generating the highest mean rewards for this user). Within each scenario we also consider two cases: (i) when users exchange full information, meaning they share the actual rewards they obtained from their choices, and (ii) when users exchange limited information, e.g., only their choices but not rewards obtained from these choices.

Algorithmic fairness in decision-making has been studied extensively in static settings where one-shot decisions are made on tasks such as classification. However, in practice most decision-making processes are of a sequential nature, where decisions made in the past may have an impact on future data. This is particularly the case when decisions affect the individuals or users generating the data used for future decisions. In this survey, we review existing literature on the fairness of data-driven sequential decision-making. We will focus on two types of sequential decisions: (1) past decisions have no impact on the underlying user population and thus no impact on future data; (2) past decisions have an impact on the underlying user population and therefore the future data, which can then impact future decisions. In each case the impact of various fairness interventions on the underlying population is examined.

In this paper we consider the problem of learning the optimal policy for uncontrolled restless bandit problems. In an uncontrolled restless bandit problem, there is a finite set of arms, each of which when pulled yields a positive reward. There is a player who sequentially selects one of the arms at each time step. The goal of the player is to maximize its undiscounted reward over a time horizon T. The reward process of each arm is a finite state Markov chain, whose transition probabilities are unknown by the player. State transitions of each arm is independent of the selection of the player. We propose a learning algorithm with logarithmic regret uniformly over time with respect to the optimal finite horizon policy. Our results extend the optimal adaptive learning of MDPs to POMDPs.

In an online contract selection problem there is a seller which offers a set of contracts to sequentially arriving buyers whose types are drawn from an unknown distribution. If there exists a profitable contract for the buyer in the offered set, i.e., a contract with payoff higher than the payoff of not accepting any contracts, the buyer chooses the contract that maximizes its payoff. In this paper we consider the online contract selection problem to maximize the sellers profit. Assuming that a structural property called ordered preferences holds for the buyer's payoff function, we propose online learning algorithms that have sub-linear regret with respect to the best set of contracts given the distribution over the buyer's type. This problem has many applications including spectrum contracts, wireless service provider data plans and recommendation systems.

In this paper, we consider the general scenario of resource sharing in a decentralized system when the resource rewards/qualities are time-varying and unknown to the users, and using the same resource by multiple users leads to reduced quality due to resource sharing. Firstly, we consider a user-independent reward model with no communication between the users, where a user gets feedback about the congestion level in the resource it uses. Secondly, we consider user-specific rewards and allow costly communication between the users. The users have a cooperative goal of achieving the highest system utility. There are multiple obstacles in achieving this goal such as the decentralized nature of the system, unknown resource qualities, communication, computation and switching costs. We propose distributed learning algorithms with logarithmic regret with respect to the optimal allocation. Our logarithmic regret result holds under both i.i.d. and Markovian reward models, as well as under communication, computation and switching costs.

We study the problem of allocating multiple users to a set of wireless channels in a decentralized manner when the channel quali- ties are time-varying and unknown to the users, and accessing the same channel by multiple users leads to reduced quality due to interference. In such a setting the users not only need to learn the inherent channel quality and at the same time the best allocations of users to channels so as to maximize the social welfare. Assuming that the users adopt a certain online learning algorithm, we investigate under what conditions the socially optimal allocation is achievable. In particular we examine the effect of different levels of knowledge the users may have and the amount of communications and cooperation. The general conclusion is that when the cooperation of users decreases and the uncertainty about channel payoffs increases it becomes harder to achieve the socially opti- mal allocation.

In this paper we study the online learning problem involving rested and restless multiarmed bandits with multiple plays. The system consists of a single player/user and a set of K finite-state discrete-time Markov chains (arms) with unknown state spaces and statistics. At each time step the player can play M arms. The objective of the user is to decide for each step which M of the K arms to play over a sequence of trials so as to maximize its long term reward. The restless multiarmed bandit is particularly relevant to the application of opportunistic spectrum access (OSA), where a (secondary) user has access to a set of K channels, each of time-varying condition as a result of random fading and/or certain primary users' activities.

We consider an opportunistic spectrum access (OSA) problem where the time-varying condition of each channel (e.g., as a result of random fading or certain primary users' activities) is modeled as an arbitrary finite-state Markov chain. At each instance of time, a (secondary) user probes a channel and collects a certain reward as a function of the state of the channel (e.g., good channel condition results in higher data rate for the user). Each channel has potentially different state space and statistics, both unknown to the user, who tries to learn which one is the best as it goes and maximizes its usage of the best channel. The objective is to construct a good online learning algorithm so as to minimize the difference between the user's performance in total rewards and that of using the best channel (on average) had it known which one is the best from a priori knowledge of the channel statistics (also known as the regret). This is a classic exploration and exploitation problem and results abound when the reward processes are assumed to be iid. Compared to prior work, the biggest difference is that in our case the reward process is assumed to be Markovian, of which iid is a special case. In addition, the reward processes are restless in that the channel conditions will continue to evolve independent of the user's actions. This leads to a restless bandit problem, for which there exists little result on either algorithms or performance bounds in this learning context to the best of our knowledge. In this paper we introduce an algorithm that utilizes regenerative cycles of a Markov chain and computes a sample-mean based index policy, and show that under mild conditions on the state transition probabilities of the Markov chains this algorithm achieves logarithmic regret uniformly over time, and that this regret bound is also optimal.

We consider the classical multi-armed bandit problem with Markovian rewards. When played an arm changes its state in a Markovian fashion while it remains frozen when not played. The player receives a state-dependent reward each time it plays an arm. The number of states and the state transition probabilities of an arm are unknown to the player. The player's objective is to maximize its long-term total reward by learning the best arm over time. We show that under certain conditions on the state transition probabilities of the arms, a sample mean based index policy achieves logarithmic regret uniformly over the total number of trials. The result shows that sample mean based index policies can be applied to learning problems under the rested Markovian bandit model without loss of optimality in the order. Moreover, comparision between Anantharam's index policy and UCB shows that by choosing a small exploration parameter UCB can have a smaller regret than Anantharam's index policy.

Combinatorial network optimization algorithms that compute optimal structures taking into account edge weights form the foundation for many network protocols. Examples include shortest path routing, minimal spanning tree computation, maximum weighted matching on bipartite graphs, etc. We present CLRMR, the first online learning algorithm that efficiently solves the stochastic version of these problems where the underlying edge weights vary as independent Markov chains with unknown dynamics. The performance of an online learning algorithm is characterized in terms of regret, defined as the cumulative difference in rewards between a suitably-defined genie, and that obtained by the given algorithm. We prove that, compared to a genie that knows the Markov transition matrices and uses the single-best structure at all times, CLRMR yields regret that is polynomial in the number of edges and nearly-logarithmic in time.

We consider a combinatorial generalization of the classical multi-armed bandit problem that is defined as follows. There is a given bipartite graph of $M$ users and $N \geq M$ resources. For each user-resource pair $(i,j)$, there is an associated state that evolves as an aperiodic irreducible finite-state Markov chain with unknown parameters, with transitions occurring each time the particular user $i$ is allocated resource $j$. The user $i$ receives a reward that depends on the corresponding state each time it is allocated the resource $j$. The system objective is to learn the best matching of users to resources so that the long-term sum of the rewards received by all users is maximized. This corresponds to minimizing regret, defined here as the gap between the expected total reward that can be obtained by the best-possible static matching and the expected total reward that can be achieved by a given algorithm. We present a polynomial-storage and polynomial-complexity-per-step matching-learning algorithm for this problem. We show that this algorithm can achieve a regret that is uniformly arbitrarily close to logarithmic in time and polynomial in the number of users and resources. This formulation is broadly applicable to scheduling and switching problems in networks and significantly extends prior results in the area.

Alternating direction method of multiplier (ADMM) is a powerful method to solve decentralized convex optimization problems. In distributed settings, each node performs computation with its local data and the local results are exchanged among neighboring nodes in an iterative fashion. During this iterative process the leakage of data privacy arises and can accumulate significantly over many iterations, making it difficult to balance the privacy-accuracy tradeoff. We propose Recycled ADMM (R-ADMM), where a linear approximation is applied to every even iteration, its solution directly calculated using only results from the previous, odd iteration. It turns out that under such a scheme, half of the updates incur no privacy loss and require much less computation compared to the conventional ADMM. Moreover, R-ADMM can be further modified (MR-ADMM) such that each node independently determines its own penalty parameter over iterations. We obtain a sufficient condition for the convergence of both algorithms and provide the privacy analysis based on objective perturbation. It can be shown that the privacy-accuracy tradeoff can be improved significantly compared with conventional ADMM.

Alternating direction method of multiplier (ADMM) is a powerful method to solve decentralized convex optimization problems. In distributed settings, each node performs computation with its local data and the local results are exchanged among neighboring nodes in an iterative fashion. During this iterative process the leakage of data privacy arises and can accumulate significantly over many iterations, making it difficult to balance the privacy-utility tradeoff. In this study we propose Recycled ADMM (R-ADMM), where a linear approximation is applied to every even iteration, its solution directly calculated using only results from the previous, odd iteration. It turns out that under such a scheme, half of the updates incur no privacy loss and require much less computation compared to the conventional ADMM. We obtain a sufficient condition for the convergence of R-ADMM and provide the privacy analysis based on objective perturbation.

Alternating direction method of multiplier (ADMM) is a popular method used to design distributed versions of a machine learning algorithm, whereby local computations are performed on local data with the output exchanged among neighbors in an iterative fashion. During this iterative process the leakage of data privacy arises. A differentially private ADMM was proposed in prior work (Zhang & Zhu, 2017) where only the privacy loss of a single node during one iteration was bounded, a method that makes it difficult to balance the tradeoff between the utility attained through distributed computation and privacy guarantees when considering the total privacy loss of all nodes over the entire iterative process. We propose a perturbation method for ADMM where the perturbed term is correlated with the penalty parameters; this is shown to improve the utility and privacy simultaneously. The method is based on a modified ADMM where each node independently determines its own penalty parameter in every iteration and decouples it from the dual updating step size. The condition for convergence of the modified ADMM and the lower bound on the convergence rate are also derived.

This paper addresses the problem of distributed learning of average belief with sequential observations, in which a network of $n>1$ agents aim to reach a consensus on the average value of their beliefs, by exchanging information only with their neighbors. Each agent has sequentially arriving samples of its belief in an online manner. The neighbor relationships among the $n$ agents are described by a graph which is possibly time-varying, whose vertices correspond to agents and whose edges depict neighbor relationships. Two distributed online algorithms are introduced for undirected and directed graphs, which are both shown to converge to the average belief almost surely. Moreover, the sequences generated by both algorithms are shown to reach consensus with an $O(1/t)$ rate with high probability, where $t$ is the number of iterations. For undirected graphs, the corresponding algorithm is modified for the case with quantized communication and limited precision of the division operation. It is shown that the modified algorithm causes all $n$ agents to either reach a quantized consensus or enter a small neighborhood around the average of their beliefs. Numerical simulations are then provided to corroborate the theoretical results.

Machine learning models trained on data from multiple demographic groups can inherit representation disparity (Hashimoto et al., 2018) that may exist in the data: the group contributing less to the training process may suffer higher loss in model accuracy; this in turn can degrade population retention in these groups over time in terms of their contribution to the training process of future models, which then exacerbates representation disparity in the long run. In this study, we seek to understand the interplay between the model accuracy and the underlying group representation and how they evolve in a sequential decision setting over an infinite horizon, and how the use of fair machine learning plays a role in this process. Using a simple user dynamics (arrival and departure) model, we characterize the long-term property of using machine learning models under a set of fairness criteria imposed on each stage of the decision process, including the commonly used statistical parity and equal opportunity fairness. We show that under this particular arrival/departure model, both these criteria cause the representation disparity to worsen over time, resulting in groups diminishing entirely from the sample pool, while the criterion of equalized loss fares much better. Our results serve to highlight the fact that fairness cannot be defined outside the larger feedback loop where past actions taken by users (who are either subject to the decisions made by the algorithm or whose data are used to train the algorithm or both) will determine future observations and decisions.

Highly expressive models such as deep neural networks (DNNs) have been widely applied to various applications and achieved increasing success. However, recent studies show that such machine learning models appear to be vulnerable against adversarial examples. So far adversarial examples have been heavily explored for 2D images, while few works have conducted to understand vulnerabilities of 3D objects which exist in real world, where 3D objects are projected to 2D domains by photo taking for different learning (recognition) tasks. In this paper, we consider adversarial behaviors in practical scenarios by manipulating the shape and texture of a given 3D mesh representation of an object. Our goal is to project the optimized "adversarial meshes" to 2D with a photorealistic renderer, and still able to mislead different machine learning models. Extensive experiments show that by generating unnoticeable 3D adversarial perturbation on shape or texture for a 3D mesh, the corresponding projected 2D instance can either lead classifiers to misclassify the victim object as an arbitrary malicious target, or hide any target object within the scene from object detectors. We conduct human studies to show that our optimized adversarial 3D perturbation is highly unnoticeable for human vision systems. In addition to the subtle perturbation for a given 3D mesh, we also propose to synthesize a realistic 3D mesh and put in a scene mimicking similar rendering conditions and therefore attack different machine learning models. In-depth analysis of transferability among various 3D renderers and vulnerable regions of meshes are provided to help better understand adversarial behaviors in real-world.

Deep Neural Networks (DNNs) have been widely applied in various recognition tasks. However, recently DNNs have been shown to be vulnerable against adversarial examples, which can mislead DNNs to make arbitrary incorrect predictions. While adversarial examples are well studied in classification tasks, other learning problems may have different properties. For instance, semantic segmentation requires additional components such as dilated convolutions and multiscale processing. In this paper, we aim to characterize adversarial examples based on spatial context information in semantic segmentation. We observe that spatial consistency information can be potentially leveraged to detect adversarial examples robustly even when a strong adaptive attacker has access to the model and detection strategies. We also show that adversarial examples based on attacks considered within the paper barely transfer among models, even though transferability is common in classification. Our observations shed new light on developing adversarial attacks and defenses to better understand the vulnerabilities of DNNs.

Deep neural networks (DNNs) have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs. Such adversarial examples can mislead DNNs to produce adversary-selected results. Different attack strategies have been proposed to generate adversarial examples, but how to produce them with high perceptual quality and more efficiently requires more research efforts. In this paper, we propose AdvGAN to generate adversarial examples with generative adversarial networks (GANs), which can learn and approximate the distribution of original instances. For AdvGAN, once the generator is trained, it can generate adversarial perturbations efficiently for any instance, so as to potentially accelerate adversarial training as defenses. We apply AdvGAN in both semi-whitebox and black-box attack settings. In semi-whitebox attacks, there is no need to access the original target model after the generator is trained, in contrast to traditional white-box attacks. In black-box attacks, we dynamically train a distilled model for the black-box model and optimize the generator accordingly. Adversarial examples generated by AdvGAN on different target models have high attack success rate under state-of-the-art defenses compared to other attacks. Our attack has placed the first with 92.76% accuracy on a public MNIST black-box attack challenge.