arXiv subjects

Li Xia

Publications and source records attributed to Li Xia.

35 records · Page 2Linked to original sources

arXiv2022

Risk-Sensitive Markov Decision Processes with Long-Run CVaR Criterion

CVaR (Conditional Value at Risk) is a risk metric widely used in finance. However, dynamically optimizing CVaR is difficult since it is not a standard Markov decision process (MDP) and the principle of dynamic programming fails. In this paper, we study the infinite-horizon discrete-time MDP with a long-run CVaR criterion, from the view of sensitivity-based optimization. By introducing a pseudo CVaR metric, we derive a CVaR difference formula which quantifies the difference of long-run CVaR under any two policies. The optimality of deterministic policies is derived. We obtain a so-called Bellman local optimality equation for CVaR, which is a necessary and sufficient condition for local optimal policies and only necessary for global optimal policies. A CVaR derivative formula is also derived for providing more sensitivity information. Then we develop a policy iteration type algorithm to efficiently optimize CVaR, which is shown to converge to local optima in the mixed policy space. We further discuss some extensions including the mean-CVaR optimization and the maximization of CVaR. Finally, we conduct numerical experiments relating to portfolio management to demonstrate the main results. Our work may shed light on dynamically optimizing CVaR from a sensitivity viewpoint.

math.OC↗

arXiv2022

A unified algorithm framework for mean-variance optimization in discounted Markov decision processes

This paper studies the risk-averse mean-variance optimization in infinite-horizon discounted Markov decision processes (MDPs). The involved variance metric concerns reward variability during the whole process, and future deviations are discounted to their present values. This discounted mean-variance optimization yields a reward function dependent on a discounted mean, and this dependency renders traditional dynamic programming methods inapplicable since it suppresses a crucial property -- time consistency. To deal with this unorthodox problem, we introduce a pseudo mean to transform the untreatable MDP to a standard one with a redefined reward function in standard form and derive a discounted mean-variance performance difference formula. With the pseudo mean, we propose a unified algorithm framework with a bilevel optimization structure for the discounted mean-variance optimization. The framework unifies a variety of algorithms for several variance-related problems including, but not limited to, risk-averse variance and mean-variance optimizations in discounted and average MDPs. Furthermore, the convergence analyses missing from the literature can be complemented with the proposed framework as well. Taking the value iteration as an example, we develop a discounted mean-variance value iteration algorithm and prove its convergence to a local optimum with the aid of a Bellman local-optimality equation. Finally, we conduct a numerical experiment on portfolio management to validate the proposed algorithm.

math.OC↗

arXiv2021

Average-Reward Reinforcement Learning with Trust Region Methods

Most of reinforcement learning algorithms optimize the discounted criterion which is beneficial to accelerate the convergence and reduce the variance of estimates. Although the discounted criterion is appropriate for certain tasks such as financial related problems, many engineering problems treat future rewards equally and prefer a long-run average criterion. In this paper, we study the reinforcement learning problem with the long-run average criterion. Firstly, we develop a unified trust region theory with discounted and average criteria and derive a novel performance bound within the trust region with the Perturbation Analysis (PA) theory. Secondly, we propose a practical algorithm named Average Policy Optimization (APO), which improves the value estimation with a novel technique named Average Value Constraint. Finally, experiments are conducted in the continuous control environment MuJoCo. In most tasks, APO performs better than the discounted PPO, which demonstrates the effectiveness of our approach. Our work provides a unified framework of the trust region approach including both the discounted and average criteria, which may complement the framework of reinforcement learning beyond the discounted objectives.

cs.LG↗

arXiv2021

Optimal Asynchronous Dynamic Policies in Energy-Efficient Data Centers

In this paper, we use a Markov decision process to find optimal asynchronous policy of an energy-efficient data center with two groups of heterogeneous servers, a finite buffer, and a fast setup process at sleep state. Servers in Group 1 always work. Servers in Group 2 may either work or sleep, and a fast setup process occurs when server's states are changed from sleep to work. In such a data center, an asynchronous dynamic policy is designed as two sub-policies: The setup policy and the sleep policy, which determine the switch rule between the work and sleep states for the servers in Group 2. To analyze the optimal asynchronous dynamic policy, we apply the Markov decision process to establish a policy-based Poisson equation, which provides expression for the unique solution of the performance potential by means of the RG-factorization. Based on this, we can characterize the monotonicity and optimality of the long-run average profit of the data center with respect to the asynchronous dynamic policy under different service prices. Furthermore, we prove that the bang-bang control is always optimal for this optimization problem, and supports a threshold-type dynamic control in the energy-efficient data center. We hope that the methodology and results derived in this paper can shed light to the study of more general energy-efficient data centers.

cs.PF↗

arXiv2021

Zero-sum risk-sensitive continuous-time stochastic games with unbounded payoff and transition rates and Borel spaces

We study a finite-horizon two-person zero-sum risk-sensitive stochastic game for continuous-time Markov chains and Borel state and action spaces, in which payoff rates, transition rates and terminal reward functions are allowed to be unbounded from below and from above and the policies can be history-dependent. Under suitable conditions, we establish the existence of a solution to the corresponding Shapley equation (SE) by an approximation technique. Then, by the SE and the extension of the Dynkin's formula, we prove the existence of a Nash equilibrium and verify that the value of the stochastic game is the unique solution to the SE. Moreover, we develop a value iteration-type algorithm for approaching to the value of the stochastic game. The convergence of the algorithm is proved by a special contraction operator in our risk-sensitive stochastic game. Finally, we demonstrate our main results by two examples.

math.OC↗

arXiv2021

Zero-Sum Semi-Markov Games with State-Action-Dependent Discount Factors

Semi-Markov model is one of the most general models for stochastic dynamic systems. This paper deals with a two-person zero-sum game for semi-Markov processes. We focus on the expected discounted payoff criterion with state-action-dependent discount factors. The state and action spaces are both Polish spaces, and the payoff function is $ω$-bounded. We first construct a fairly general model of semi-Markov games under a given semi-Markov kernel and a pair of strategies. Next, based on the standard regularity condition and the continuity-compactness condition for semi-Markov games, we derive a "drift condition" on the semi-Markov kernel and suppose that the discount factors have a positive lower bound, under which the existence of the value function and a pair of optimal stationary strategies of our semi-Markov game are proved by using the Shapley equation. Moreover, when the state and action spaces are both finite, a value iteration-type algorithm for computing the value function and $\varepsilon$-Nash equilibrium of the game is developed. The convergence of the algorithm is also proved. Finally, we conduct numerical examples to demonstrate our main results.

cs.GT↗

arXiv2020

Wasserstein Distance guided Adversarial Imitation Learning with Reward Shape Exploration

The generative adversarial imitation learning (GAIL) has provided an adversarial learning framework for imitating expert policy from demonstrations in high-dimensional continuous tasks. However, almost all GAIL and its extensions only design a kind of reward function of logarithmic form in the adversarial training strategy with the Jensen-Shannon (JS) divergence for all complex environments. The fixed logarithmic type of reward function may be difficult to solve all complex tasks, and the vanishing gradients problem caused by the JS divergence will harm the adversarial learning process. In this paper, we propose a new algorithm named Wasserstein Distance guided Adversarial Imitation Learning (WDAIL) for promoting the performance of imitation learning (IL). There are three improvements in our method: (a) introducing the Wasserstein distance to obtain more appropriate measure in the adversarial training process, (b) using proximal policy optimization (PPO) in the reinforcement learning stage which is much simpler to implement and makes the algorithm more efficient, and (c) exploring different reward function shapes to suit different tasks for improving the performance. The experiment results show that the learning procedure remains remarkably stable, and achieves significant performance in the complex continuous control tasks of MuJoCo.

cs.LG↗

arXiv2020

Risk-Sensitive Markov Decision Processes with Combined Metrics of Mean and Variance

This paper investigates the optimization problem of an infinite stage discrete time Markov decision process (MDP) with a long-run average metric considering both mean and variance of rewards together. Such performance metric is important since the mean indicates average returns and the variance indicates risk or fairness. However, the variance metric couples the rewards at all stages, the traditional dynamic programming is inapplicable as the principle of time consistency fails. We study this problem from a new perspective called the sensitivity-based optimization theory. A performance difference formula is derived and it can quantify the difference of the mean-variance combined metrics of MDPs under any two different policies. The difference formula can be utilized to generate new policies with strictly improved mean-variance performance. A necessary condition of the optimal policy and the optimality of deterministic policies are derived. We further develop an iterative algorithm with a form of policy iteration, which is proved to converge to local optima both in the mixed and randomized policy space. Specially, when the mean reward is constant in policies, the algorithm is guaranteed to converge to the global optimum. Finally, we apply our approach to study the fluctuation reduction of wind power in an energy storage system, which demonstrates the potential applicability of our optimization method.

math.OC↗

arXiv2020

Embedding-based Retrieval in Facebook Search

Search in social networks such as Facebook poses different challenges than in classical web search: besides the query text, it is important to take into account the searcher's context to provide relevant results. Their social graph is an integral part of this context and is a unique aspect of Facebook search. While embedding-based retrieval (EBR) has been applied in eb search engines for years, Facebook search was still mainly based on a Boolean matching model. In this paper, we discuss the techniques for applying EBR to a Facebook Search system. We introduce the unified embedding framework developed to model semantic embeddings for personalized search, and the system to serve embedding-based retrieval in a typical search system based on an inverted index. We discuss various tricks and experiences on end-to-end optimization of the whole system, including ANN parameter tuning and full-stack optimization. Finally, we present our progress on two selected advanced topics about modeling. We evaluated EBR on verticals for Facebook Search with significant metrics gains observed in online A/B experiments. We believe this paper will provide useful insights and experiences to help people on developing embedding-based retrieval systems in search engines.

cs.IR↗

arXiv2020

On the existence of optimal stationary policies for average Markov decision processes with countable states

For a Markov decision process with countably infinite states, the optimal value may not be achievable in the set of stationary policies. In this paper, we study the existence conditions of an optimal stationary policy in a countable-state Markov decision process under the long-run average criterion. With a properly defined metric on the policy space of ergodic MDPs, the existence of an optimal stationary policy can be guaranteed by the compactness of the space and the continuity of the long-run average cost with respect to the metric. We further extend this condition by some assumptions which can be easily verified in control problems of specific systems, such as queueing systems. Our results make a complementary contribution to the literature in the sense that our method is capable to handle the cost function unbounded from both below and above, only at the condition of continuity and ergodicity. Several examples are provided to illustrate the application of our main results.

math.OC↗

arXiv2020

SOAC: The Soft Option Actor-Critic Architecture

The option framework has shown great promise by automatically extracting temporally-extended sub-tasks from a long-horizon task. Methods have been proposed for concurrently learning low-level intra-option policies and high-level option selection policy. However, existing methods typically suffer from two major challenges: ineffective exploration and unstable updates. In this paper, we present a novel and stable off-policy approach that builds on the maximum entropy model to address these challenges. Our approach introduces an information-theoretical intrinsic reward for encouraging the identification of diverse and effective options. Meanwhile, we utilize a probability inference model to simplify the optimization problem as fitting optimal trajectories. Experimental results demonstrate that our approach significantly outperforms prior on-policy and off-policy methods in a range of Mujoco benchmark tasks while still providing benefits for transfer learning. In these tasks, our approach learns a diverse set of options, each of whose state-action space has strong coherence.

cs.AI↗

arXiv2019

An Overview for Markov Decision Processes in Queues and Networks

Markov decision processes (MDPs) in queues and networks have been an interesting topic in many practical areas since the 1960s. This paper provides a detailed overview on this topic and tracks the evolution of many basic results. Also, this paper summarizes several interesting directions in the future research. We hope that this overview can shed light to MDPs in queues and networks, and also to their extensive applications in various practical areas.

math.OC↗

arXiv2018

A $c/μ$-Rule for Service Resource Allocation in Group-Server Queues

In this paper, we study a dynamic on/off server scheduling problem in a queueing system with multi-class servers, where servers are heterogeneous and can be classified into $K$ groups. Servers in the same group are homogeneous. A scheduling policy determines the number of working servers (servers that are turned on) in each group at every state $n$ (number of customers in the system). Our goal is to find the optimal scheduling policy to minimize the long-run average cost, which consists of an increasing convex holding cost and a linear operating cost. We use the sensitivity-based optimization theory to characterize the optimal policy. A necessary and sufficient condition of the optimal policy is derived. We also prove that the optimal policy has monotone structures and a quasi bang-bang control is optimal. We find that the optimal policy is indexed by the value of $c - μG(n)$, where $c$ is the operating cost rate, $μ$ is the service rate for a server, and $G(n)$ is a computable quantity called perturbation realization factor. Specifically, the group with smaller negative $c - μG(n)$ is more preferred to be turned on, while the group with positive $c - μG(n)$ should be turned off. However, the preference ranking of each group is affected by $G(n)$ and the preference order may change with the state $n$, the arrival rate, and the cost function. Under a reasonable condition of scale economies, we further prove that the optimal policy obeys a so-called $c$/$μ$-rule. That is, the servers with smaller $c$/$μ$ should be turned on with higher priority and the preference order of groups remains unchanged. This rule can be viewed as a sister version of the famous $cμ$-rule for polling queues. With the monotone property of $G(n)$, we further prove that the optimal policy has a multi-threshold structure when the $c$/$μ$-rule is applied.

math.OC↗

arXiv2018

Optimal Energy-Efficient Policies for Data Centers through Sensitivity-Based Optimization

In this paper, we propose a novel dynamic decision method by applying the sensitivity-based optimization theory to find the optimal energy-efficient policy of a data center with two groups of heterogeneous servers. Servers in Group 1 always work at high energy consumption, while servers in Group 2 may either work at high energy consumption or sleep at low energy consumption. An energy-efficient control policy determines the switch between work and sleep states of servers in Group 2 in a dynamic way. Since servers in Group 1 are always working with high priority to jobs, a transfer rule is proposed to migrate the jobs in Group 2 to idle servers in Group 1. To find the optimal energy-efficient policy, we set up a policy-based Poisson equation, and provide explicit expressions for its unique solution of performance potentials by means of the RG-factorization. Based on this, we characterize monotonicity and optimality of the long-run average profit with respect to the policies under different service prices. We prove that the bang-bang control is always optimal for this optimization problem, i.e., we should either keep all servers sleep or turn on the servers such that the number of working servers equals that of waiting jobs in Group 2. As an easy adoption of policy forms, we further study the threshold-type policy and obtain a necessary condition of the optimal threshold policy. We hope the methodology and results derived in this paper can shed light to the study of more general energy-efficient data centers.

math.OC↗

arXiv2017

Mean-Variance Optimization of Discrete Time Discounted Markov Decision Processes

In this paper, we study a mean-variance optimization problem in an infinite horizon discrete time discounted Markov decision process (MDP). The objective is to minimize the variance of system rewards with the constraint of mean performance. Different from most of works in the literature which require the mean performance already achieve optimum, we can let the mean discounted performance equal any constant. The difficulty of this problem is caused by the quadratic form of the variance function which makes the variance minimization problem not a standard MDP. By proving the decomposable structure of the feasible policy space, we transform this constrained variance minimization problem to an equivalent unconstrained MDP under a new discounted criterion and a new reward function. The difference of the variances of Markov chains under any two feasible policies is quantified by a difference formula. Based on the variance difference formula, a policy iteration algorithm is developed to find the optimal policy. We also prove the optimality of deterministic policy over the randomized policy generated in the mean-constrained policy space. Numerical experiments demonstrate the effectiveness of our approach.

math.OC↗

arXiv2017

Group-Server Queues

By analyzing energy-efficient management of data centers, this paper proposes and develops a class of interesting {\it Group-Server Queues}, and establishes two representative group-server queues through loss networks and impatient customers, respectively. Furthermore, such two group-server queues are given model descriptions and necessary interpretation. Also, simple mathematical discussion is provided, and simulations are made to study the expected queue lengths, the expected sojourn times and the expected virtual service times. In addition, this paper also shows that this class of group-server queues are often encountered in many other practical areas including communication networks, manufacturing systems, transportation networks, financial networks and healthcare systems. Note that the group-server queues are always used to design effectively dynamic control mechanisms through regrouping and recombining such many servers in a large-scale service system by means of, for example, bilateral threshold control, and customers transfer to the buffer or server groups. This leads to the large-scale service system that is divided into several adaptive and self-organizing subsystems through scheduling of batch customers and regrouping of service resources, which make the middle layer of this service system more effectively managed and strengthened under a dynamic, real-time and even reward optimal framework. Based on this, performance of such a large-scale service system may be improved greatly in terms of introducing and analyzing such group-server queues. Therefore, not only analysis of group-server queues is regarded as a new interesting research direction, but there also exists many theoretical challenges, basic difficulties and open problems in the area of queueing networks.

cs.PF↗

arXiv2016

A Generalized Fundamental Matrix for Computing Fundamental Quantities of Markov Systems

As is well known, the fundamental matrix $(I - P + e π)^{-1}$ plays an important role in the performance analysis of Markov systems, where $P$ is the transition probability matrix, $e$ is the column vector of ones, and $π$ is the row vector of the steady state distribution. It is used to compute the performance potential (relative value function) of Markov decision processes under the average criterion, such as $g=(I - P + e π)^{-1} f$ where $g$ is the column vector of performance potentials and $f$ is the column vector of reward functions. However, we need to pre-compute $π$ before we can compute $(I - P + e π)^{-1}$. In this paper, we derive a generalization version of the fundamental matrix as $(I - P + e r)^{-1}$, where $r$ can be any given row vector satisfying $r e \neq 0$. With this generalized fundamental matrix, we can compute $g=(I - P + e r)^{-1} f$. The steady state distribution is computed as $π= r(I - P + e r)^{-1}$. The Q-factors at every state-action pair can also be computed in a similar way. These formulas may give some insights on further understanding how to efficiently compute or estimate the values of $g$, $π$, and Q-factors in Markov systems, which are fundamental quantities for the performance optimization of Markov systems.

math.OC↗