arXiv subjects

Kuang Xu

Publications and source records attributed to Kuang Xu.

At least 19 recordsLinked to original sources

arXiv2025

TeleEval-OS: Performance evaluations of large language models for operations scheduling

The rapid advancement of large language models (LLMs) has significantly propelled progress in artificial intelligence, demonstrating substantial application potential across multiple specialized domains. Telecommunications operation scheduling (OS) is a critical aspect of the telecommunications industry, involving the coordinated management of networks, services, risks, and human resources to optimize production scheduling and ensure unified service control. However, the inherent complexity and domain-specific nature of OS tasks, coupled with the absence of comprehensive evaluation benchmarks, have hindered thorough exploration of LLMs' application potential in this critical field. To address this research gap, we propose the first Telecommunications Operation Scheduling Evaluation Benchmark (TeleEval-OS). Specifically, this benchmark comprises 15 datasets across 13 subtasks, comprehensively simulating four key operational stages: intelligent ticket creation, intelligent ticket handling, intelligent ticket closure, and intelligent evaluation. To systematically assess the performance of LLMs on tasks of varying complexity, we categorize their capabilities in telecommunications operation scheduling into four hierarchical levels, arranged in ascending order of difficulty: basic NLP, knowledge Q&A, report generation, and report analysis. On TeleEval-OS, we leverage zero-shot and few-shot evaluation methods to comprehensively assess 10 open-source LLMs (e.g., DeepSeek-V3) and 4 closed-source LLMs (e.g., GPT-4o) across diverse scenarios. Experimental results demonstrate that open-source LLMs can outperform closed-source LLMs in specific scenarios, highlighting their significant potential and value in the field of telecommunications operation scheduling.

cs.CL↗

arXiv2022

Temporal Concatenation for Markov Decision Processes

We propose and analyze a temporal concatenation heuristic for solving large-scale finite-horizon Markov decision processes (MDP), which divides the MDP into smaller sub-problems along the time horizon and generates an overall solution by simply concatenating the optimal solutions from these sub-problems. As a "black box" architecture, temporal concatenation works with a wide range of existing MDP algorithms. Our main results characterize the regret of temporal concatenation compared to the optimal solution. We provide upper bounds for general MDP instances, as well as a family of MDP instances in which the upper bounds are shown to be tight. Together, our results demonstrate temporal concatenation's potential of substantial speed-up at the expense of some performance degradation.

math.OC↗

arXiv2022

Gaussian Imagination in Bandit Learning

Assuming distributions are Gaussian often facilitates computations that are otherwise intractable. We study the performance of an agent that attains a bounded information ratio with respect to a bandit environment with a Gaussian prior distribution and a Gaussian likelihood function when applied instead to a Bernoulli bandit. Relative to an information-theoretic bound on the Bayesian regret the agent would incur when interacting with the Gaussian bandit, we bound the increase in regret when the agent interacts with the Bernoulli bandit. If the Gaussian prior distribution and likelihood function are sufficiently diffuse, this increase grows at a rate which is at most linear in the square-root of the time horizon, and thus the per-timestep increase vanishes. Our results formalize the folklore that so-called Bayesian agents remain effective when instantiated with diffuse misspecified distributions.

cs.LG↗

arXiv2021

Learner-Private Convex Optimization

Convex optimization with feedback is a framework where a learner relies on iterative queries and feedback to arrive at the minimizer of a convex function. It has gained considerable popularity thanks to its scalability in large-scale optimization and machine learning. The repeated interactions, however, expose the learner to privacy risks from eavesdropping adversaries that observe the submitted queries. In this paper, we study how to optimally obfuscate the learner's queries in convex optimization with first-order feedback, so that their learned optimal value is provably difficult to estimate for an eavesdropping adversary. We consider two formulations of learner privacy: a Bayesian formulation in which the convex function is drawn randomly, and a minimax formulation in which the function is fixed and the adversary's probability of error is measured with respect to a minimax criterion. Suppose that the learner wishes to ensure the adversary cannot estimate accurately with probability greater than $1/L$ for some $L>0$. Our main results show that the query complexity overhead is additive in $L$ in the minimax formulation, but multiplicative in $L$ in the Bayesian formulation. Compared to existing learner-private sequential learning models with binary feedback, our results apply to the significantly richer family of general convex functions with full-gradient feedback. Our proofs learn on tools from the theory of Dirichlet processes, as well as a novel strategy designed for measuring information leakage under a full-gradient oracle.

stat.ML↗

arXiv2021

Learning and Information in Stochastic Networks and Queues

We review the role of information and learning in the stability and optimization of queueing systems. In recent years, techniques from supervised learning, bandit learning and reinforcement learning have been applied to queueing systems supported by increasing role of information in decision making. We present observations and new results that help rationalize the application of these areas to queueing systems. We prove that the MaxWeight and BackPressure policies are an application of Blackwell's Approachability Theorem. This connects queueing theoretic results with adversarial learning. We then discuss the requirements of statistical learning for service parameter estimation. As an example, we show how queue size regret can be bounded when applying a perceptron algorithm to classify service. Next, we discuss the role of state information in improved decision making. Here we contrast the roles of epistemic information (information on uncertain parameters) and aleatoric information (information on an uncertain state). Finally we review recent advances in the theory of reinforcement learning and queueing, as well as, provide discussion on current research challenges.

cs.LG↗

arXiv2021

Hierarchical Causal Bandit

Causal bandit is a nascent learning model where an agent sequentially experiments in a causal network of variables, in order to identify the reward-maximizing intervention. Despite the model's wide applicability, existing analytical results are largely restricted to a parallel bandit version where all variables are mutually independent. We introduce in this work the hierarchical causal bandit model as a viable path towards understanding general causal bandits with dependent variables. The core idea is to incorporate a contextual variable that captures the interaction among all variables with direct effects. Using this hierarchical framework, we derive sharp insights on algorithmic design in causal bandits with dependent arms and obtain nearly matching regret bounds in the case of a binary context.

stat.ML↗

arXiv2021

A Bit Better? Quantifying Information for Bandit Learning

The information ratio offers an approach to assessing the efficacy with which an agent balances between exploration and exploitation. Originally, this was defined to be the ratio between squared expected regret and the mutual information between the environment and action-observation pair, which represents a measure of information gain. Recent work has inspired consideration of alternative information measures, particularly for use in analysis of bandit learning algorithms to arrive at tighter regret bounds. We investigate whether quantification of information via such alternatives can improve the realized performance of information-directed sampling, which aims to minimize the information ratio.

cs.LG↗

arXiv2020

Anonymous Stochastic Routing

We propose and analyze a recipient-anonymous stochastic routing model to study a fundamental trade-off between anonymity and routing delay. An agent wants to quickly reach a goal vertex in a network through a sequence of routing actions, while an overseeing adversary observes the agent's entire trajectory and tries to identify her goal among those vertices traversed. We are interested in understanding the probability that the adversary can correctly identify the agent's goal (anonymity), as a function of the time it takes the agent to reach it (delay). A key feature of our model is the presence of intrinsic uncertainty in the environment, so that each of the agent's intended steps is subject to random perturbation and thus may not materialize as planned. Using large-network asymptotics, our main results provide near-optimal characterization of the anonymity-delay trade-off under a number of network topologies. Our main technical contributions are centered around a new class of "noise-harnessing" routing strategies that adaptively combine intrinsic uncertainty from the environment with additional artificial randomization to achieve provably efficient obfuscation.

cs.GT↗

arXiv2020

Optimal query complexity for private sequential learning against eavesdropping

We study the query complexity of a learner-private sequential learning problem, motivated by the privacy and security concerns due to eavesdropping that arise in practical applications such as pricing and Federated Learning. A learner tries to estimate an unknown scalar value, by sequentially querying an external database and receiving binary responses; meanwhile, a third-party adversary observes the learner's queries but not the responses. The learner's goal is to design a querying strategy with the minimum number of queries (optimal query complexity) so that she can accurately estimate the true value, while the eavesdropping adversary even with the complete knowledge of her querying strategy cannot. We develop new querying strategies and analytical techniques and use them to prove tight upper and lower bounds on the optimal query complexity. The bounds almost match across the entire parameter range, substantially improving upon existing results. We thus obtain a complete picture of the optimal query complexity as a function of the estimation accuracy and the desired levels of privacy. We also extend the results to sequential learning models in higher dimensions, and where the binary responses are noisy. Our analysis leverages a crucial insight into the nature of private learning problem, which suggests that the query trajectory of an optimal learner can be divided into distinct phases that focus on pure learning versus learning and obfuscation, respectively.

stat.ML↗

arXiv2020

Experimenting in Equilibrium

Classical approaches to experimental design assume that intervening on one unit does not affect other units. There are many important settings, however, where this non-interference assumption does not hold, as when running experiments on supply-side incentives on a ride-sharing platform or subsidies in an energy marketplace. In this paper, we introduce a new approach to experimental design in large-scale stochastic systems with considerable cross-unit interference, under an assumption that the interference is structured enough that it can be captured via mean-field modeling. Our approach enables us to accurately estimate the effect of small changes to system parameters by combining unobstrusive randomization with lightweight modeling, all while remaining in equilibrium. We can then use these estimates to optimize the system by gradient descent. Concretely, we focus on the problem of a platform that seeks to optimize supply-side payments p in a centralized marketplace where different suppliers interact via their effects on the overall supply-demand equilibrium, and show that our approach enables the platform to optimize p in large systems using vanishingly small perturbations.

math.OC↗

arXiv2020

Private Sequential Learning

We formulate a private learning model to study an intrinsic tradeoff between privacy and query complexity in sequential learning. Our model involves a learner who aims to determine a scalar value, $v^*$, by sequentially querying an external database and receiving binary responses. In the meantime, an adversary observes the learner's queries, though not the responses, and tries to infer from them the value of $v^*$. The objective of the learner is to obtain an accurate estimate of $v^*$ using only a small number of queries, while simultaneously protecting her privacy by making $v^*$ provably difficult to learn for the adversary. Our main results provide tight upper and lower bounds on the learner's query complexity as a function of desired levels of privacy and estimation accuracy. We also construct explicit query strategies whose complexity is optimal up to an additive constant.

cs.LG↗

arXiv2019

Query Complexity of Bayesian Private Learning

We study the query complexity of Bayesian Private Learning: a learner wishes to locate a random target within an interval by submitting queries, in the presence of an adversary who observes all of her queries but not the responses. How many queries are necessary and sufficient in order for the learner to accurately estimate the target, while simultaneously concealing the target from the adversary? Our main result is a query complexity lower bound that is tight up to the first order. We show that if the learner wants to estimate the target within an error of $\varepsilon$, while ensuring that no adversary estimator can achieve a constant additive error with probability greater than $1/L$, then the query complexity is on the order of $L\log(1/\varepsilon)$, as $\varepsilon \to 0$. Our result demonstrates that increased privacy, as captured by $L$, comes at the expense of a {multiplicative} increase in query complexity. Our proof method builds on Fano's inequality and a family of proportional-sampling estimators. As an illustration of the method's wider applicability, we generalize the complexity lower bound to settings involving high-dimensional linear query learning and partial adversary observation.

cs.LG↗

arXiv2019

Reinforcement with Fading Memories

We study the effect of imperfect memory on decision making in the context of a stochastic sequential action-reward problem. An agent chooses a sequence of actions which generate discrete rewards at different rates. She is allowed to make new choices at rate $β$, while past rewards disappear from her memory at rate $μ$. We focus on a family of decision rules where the agent makes a new choice by randomly selecting an action with a probability approximately proportional to the amount of past rewards associated with each action in her memory. We provide closed-form formulae for the agent's steady-state choice distribution in the regime where the memory span is large ($μ\to 0$), and show that the agent's success critically depends on how quickly she updates her choices relative to the speed of memory decay. If $β\gg μ$, the agent almost always chooses the best action, i.e., the one with the highest reward rate. Conversely, if $β\ll μ$, the agent chooses an action with a probability roughly proportional to its reward rate.

math.PR↗

arXiv2019

Information and Memory in Dynamic Resource Allocation

We propose a general framework, dubbed Stochastic Processing under Imperfect Information (SPII), to study the impact of information constraints and memories on dynamic resource allocation. The framework involves a Stochastic Processing Network (SPN) scheduling problem in which the scheduler may access the system state only through a noisy channel, and resource allocation decisions must be carried out through the interaction between an encoding policy (who observes the state) and allocation policy (who chooses the allocation). Applications in the management of large-scale data centers and human-in-the-loop service systems are among our chief motivations. We quantify the degree to which information constraints reduce the size of the capacity region in general SPNs, and how such reduction depends on the amount of memories available to the encoding and allocation policies. Using a novel metric, capacity factor, our main theorem characterizes the reduction in capacity region (under "optimal" policies) for all non-degenerate channels, and across almost all combinations of memory sizes. Notably, the theorem demonstrates, in substantial generality, that (1) the presence of a noisy channel always reduces capacity, (2) more memory for the allocation policy always improves capacity, and (3) more memory for the encoding policy has little to no effect on capacity. Finally, all of our positive (achievability) results are established through constructive, implementable policies.

cs.IT↗

arXiv2017

Flexible Queueing Architectures

We study a multi-server model with $n$ flexible servers and $n$ queues, connected through a bipartite graph, where the level of flexibility is captured by the graph's average degree, $d_n$. Applications in content replication in data centers, skill-based routing in call centers, and flexible supply chains are among our main motivations. We focus on the scaling regime where the system size $n$ tends to infinity, while the overall traffic intensity stays fixed. We show that a large capacity region and an asymptotically vanishing queueing delay are simultaneously achievable even under limited flexibility ($d_n \ll n$). Our main results demonstrate that, when $d_n\gg \ln n$, a family of expander-graph-based flexibility architectures has a capacity region that is within a constant factor of the maximum possible, while simultaneously ensuring a diminishing queueing delay for all arrival rate vectors in the capacity region. Our analysis is centered around a new class of virtual-queue-based scheduling policies that rely on dynamically constructed job-to-server assignments on the connectivity graph. For comparison, we also analyze a natural family of modular architectures, which is simpler but has provably weaker performance.

math.PR↗

arXiv2016

On the capacity of information processing systems

We propose and analyze a family of information processing systems, where a finite set of experts or servers are employed to extract information about a stream of incoming jobs. Each job is associated with a hidden label drawn from some prior distribution. An inspection by an expert produces a noisy outcome that depends both on the job's hidden label and the type of the expert, and occupies the expert for a finite time duration. A decision maker's task is to dynamically assign inspections so that the resulting outcomes can be used to accurately recover the labels of all jobs, while keeping the system stable. Among our chief motivations are applications in crowd-sourcing, diagnostics, and experiment designs, where one wishes to efficiently learn the nature of a large number of items, using a finite pool of computational resources or human agents. We focus on the capacity of such an information processing system. Given a level of accuracy guarantee, we ask how many experts are needed in order to stabilize the system, and through what inspection architecture. Our main result provides an adaptive inspection policy that is asymptotically optimal in the following sense: the ratio between the required number of experts under our policy and the theoretical optimal converges to one, as the probability of error in label recovery tends to zero.

math.PR↗

arXiv2015

Necessity of Future Information in Admission Control

We study the necessity of predictive information in a class of queueing admission control problems, where a system manager is allowed to divert incoming jobs up to a fixed rate, in order to minimize the queueing delay experienced by the admitted jobs. Spencer et al. (2014) show that the system's delay performance can be significantly improved by having access to future information in the form of a lookahead window, during which the times of future arrivals and services are revealed. They prove that, while delay under an optimal online policy diverges to infinity in the heavy-traffic regime, it can stay bounded by making use of future information. However, the diversion polices of Spencer et al. (2014) require the length of the lookahead window to grow to infinity at a non-trivial rate in the heavy-traffic regime, and it remained open whether substantial performance improvement could still be achieved with less future information. We resolve this question to a large extent by establishing an asymptotically tight lower bound on how much future information is necessary to achieve superior performance, which matches the upper bound of Spencer et al. (2014) up to a constant multiplicative factor. Our result hence demonstrates that the system's heavy-traffic delay performance is highly sensitive to the amount of future information available. Our proof is based on analyzing certain excursion probabilities of the input sample paths, and exploiting a connection between a policy's diversion decisions and subsequent server idling, which may be of independent interest for related dynamic resource allocation problems.

math.PR↗

arXiv2014

Queuing with future information

We study an admissions control problem, where a queue with service rate $1-p$ receives incoming jobs at rate $λ\in(1-p,1)$, and the decision maker is allowed to redirect away jobs up to a rate of $p$, with the objective of minimizing the time-average queue length. We show that the amount of information about the future has a significant impact on system performance, in the heavy-traffic regime. When the future is unknown, the optimal average queue length diverges at rate $\sim\log_{1/(1-p)}\frac{1}{1-λ}$, as $λ\to 1$. In sharp contrast, when all future arrival and service times are revealed beforehand, the optimal average queue length converges to a finite constant, $(1-p)/p$, as $λ\to1$. We further show that the finite limit of $(1-p)/p$ can be achieved using only a finite lookahead window starting from the current time frame, whose length scales as $\mathcal{O}(\log\frac{1}{1-λ})$, as $λ\to1$. This leads to the conjecture of an interesting duality between queuing delay and the amount of information about the future.

math.PR↗