Xiangxiang Dai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Xiangxiang Dai has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- John Chi-Shing Lui
The Chinese University of Hong Kong (CUHK)
15 shared publications - Xutong Liu
University of Washington, Tacoma, USA
7 shared publications - Peng Yang
Huazhong University of Science and Technology
6 shared publications - Mohammad Hajiesmaili
University of Massachusetts Amherst, Amherst, USA
3 shared publications - Xuchuang Wang
Hong Kong Baptisit University, Hong Kong, Hong Kong
3 shared publications Xutong Liu
Department of Computer Science and Systems, University of Washington, Tacoma, WA, USA
3 shared publications
Latest Publications
- 5 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. PP (15361233) , 1-15
Online Outcome-Driven Configuration Selection With Adaptive Grouping for Live Video Streams
Optimizing live video stream configurations, such as resolution, is essential for balancing inference accuracy and bitrate, thereby reducing data volume and conserving network resources in intelligent video analytics. Existing systems predominantly rely on offline or periodic strategies using historical data to determine optimal configurations. However, these methods may fall short in managing bitrate, which is highly sensitive to dynamic variations in video content and can benefit from real-time adjustments. To address these limitations, we propose the O$_{2}$ MEGA (Online Outcome-driven Grouping Configuration Adaptation) framework. Motivated by unique experimental observations, O$_{2}$ MEGA adapts configurations in an “outcome-driven” manner by directly utilizing real-time online bitrate outcomes from encoded video streams. This approach allows efficient and lightweight integration of outcome feedback into configuration decisions without incurring significant analysis overhead that could negate the benefits of adaptation. Moreover, O$_{2}$ MEGA introduces a dual grouping mechanism that enables cross-camera collaboration on configuration sharing, expediting configuration selection in densely deployed camera networks. Unlike existing methods that rely solely on spatio-temporal correlations, our framework groups cameras based on real-time outcome feedback, thus adaptively optimizing group-level performance online. We provide theoretical guarantees to ensure the interpretability and effectiveness of the framework. Experimental results on a testbed with four NVIDIA development kits and an edge server demonstrate up to 6.8% accuracy improvement or 13.2% data volume reduction compared to the state-of-the-art under four typical vision queries.- 5 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. PP (15361233) , 1-15
Online Outcome-Driven Configuration Selection With Adaptive Grouping for Live Video Streams
Optimizing live video stream configurations, such as resolution, is essential for balancing inference accuracy and bitrate, thereby reducing data volume and conserving network resources in intelligent video analytics. Existing systems predominantly rely on offline or periodic strategies using historical data to determine optimal configurations. However, these methods may fall short in managing bitrate, which is highly sensitive to dynamic variations in video content and can benefit from real-time adjustments. To address these limitations, we propose the O$_{2}$ MEGA (Online Outcome-driven Grouping Configuration Adaptation) framework. Motivated by unique experimental observations, O$_{2}$ MEGA adapts configurations in an “outcome-driven” manner by directly utilizing real-time online bitrate outcomes from encoded video streams. This approach allows efficient and lightweight integration of outcome feedback into configuration decisions without incurring significant analysis overhead that could negate the benefits of adaptation. Moreover, O$_{2}$ MEGA introduces a dual grouping mechanism that enables cross-camera collaboration on configuration sharing, expediting configuration selection in densely deployed camera networks. Unlike existing methods that rely solely on spatio-temporal correlations, our framework groups cameras based on real-time outcome feedback, thus adaptively optimizing group-level performance online. We provide theoretical guarantees to ensure the interpretability and effectiveness of the framework. Experimental results on a testbed with four NVIDIA development kits and an edge server demonstrate up to 6.8% accuracy improvement or 13.2% data volume reduction compared to the state-of-the-art under four typical vision queries.- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2183-2194
BANCO: Drift-Aware Batched Bandits for Adaptive Proximity Graph Pruning
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 2183-2194
Open AccessBANCO: Drift-Aware Batched Bandits for Adaptive Proximity Graph Pruning
- 27 February 2026
- Published by Elsevier in Artificial Intelligence
- Vol. 354
Exploring multi-layered networks through random walks: bridging offline optimization and online learning
- 27 February 2026
- Published by Elsevier in Artificial Intelligence
- Vol. 354
Exploring multi-layered networks through random walks: bridging offline optimization and online learning
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 3932-3946
Combinatorial Logistic Online Learning and Its Applications in Nonlinear Networked Systems
Combinatorial multi-armed bandit (CMAB) is a fundamental online learning framework that can optimize cumulative rewards in networked systems under uncertainty. Real-world applications like content delivery and channel allocation often feature binary base arm rewards and nonlinear total reward functions. This paper introduces combinatorial logistic bandits (CLogB), a contextual CMAB framework with the base arm reward modeled as a nonlinear logistic function of the context, and the feedback is governed by a general arm-triggering process. We study CLogB with smooth reward functions, covering applications such as online content delivery, online multi-LLM selection, and dynamic channel allocation. Our first algorithm, CLogUCB, uses a variance-agnostic exploration bonus and achieves a regret bound of Õ(d√κKT), where d is the feature dimension, κ reflects logistic model nonlinearity, K is the maximum number of triggered arms, and Õ ignores logarithmic factors. This improves on prior results by Õ (√κ). We further propose VA-CLogUCB, a variance-adaptive enhancement achieving regret bounds of Õ(d√KT) under standard smoothness conditions and Õ (d√T) under stronger variance conditions, removing dependence on K. For time-invariant feature maps, we enhance computational efficiency by avoiding nonconvex optimization while maintaining Õ(d√T) regret. Experiments on synthetic and real-world datasets validate the superior performance of our algorithms, demonstrating their effectiveness and scalability for real-world networked systems.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 3932-3946
Combinatorial Logistic Online Learning and Its Applications in Nonlinear Networked Systems
Combinatorial multi-armed bandit (CMAB) is a fundamental online learning framework that can optimize cumulative rewards in networked systems under uncertainty. Real-world applications like content delivery and channel allocation often feature binary base arm rewards and nonlinear total reward functions. This paper introduces combinatorial logistic bandits (CLogB), a contextual CMAB framework with the base arm reward modeled as a nonlinear logistic function of the context, and the feedback is governed by a general arm-triggering process. We study CLogB with smooth reward functions, covering applications such as online content delivery, online multi-LLM selection, and dynamic channel allocation. Our first algorithm, CLogUCB, uses a variance-agnostic exploration bonus and achieves a regret bound of Õ(d√κKT), where d is the feature dimension, κ reflects logistic model nonlinearity, K is the maximum number of triggered arms, and Õ ignores logarithmic factors. This improves on prior results by Õ (√κ). We further propose VA-CLogUCB, a variance-adaptive enhancement achieving regret bounds of Õ(d√KT) under standard smoothness conditions and Õ (d√T) under stronger variance conditions, removing dependence on K. For time-invariant feature maps, we enhance computational efficiency by avoiding nonconvex optimization while maintaining Õ(d√T) regret. Experiments on synthetic and real-world datasets validate the superior performance of our algorithms, demonstrating their effectiveness and scalability for real-world networked systems.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 44, 6344-6359
Communication-Efficient Distributed Learning for Perspective-Aware Visual Inference Over Edge-Cloud Networks
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 44, 6344-6359
Open AccessCommunication-Efficient Distributed Learning for Perspective-Aware Visual Inference Over Edge-Cloud Networks
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 6648-6663
Online Minimization of Convex Age of Information With Transmission Costs
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 6648-6663
Online Minimization of Convex Age of Information With Transmission Costs
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 12, 5063-5076
Networked Edge Resource Orchestration for Mobile AI-Generated Content Services
In this paper, we design an edge-assisted mobile Artificial Intelligence-Generated Content (AIGC) service system. The core of this system is the orchestrated multi-dimensional resource allocation and generative AI model deployment for efficient AIGC service provisioning at the edge. The heterogeneous resource demands of these models strain scarce resources of edge servers. This scarcity calls for a resource-efficient allocation policy whose goal conflicts with ensuring resource fairness for tasks requested by concurrent users. Specifically, such a policy risks enabling resource monopolization by certain tasks, starving others. To address these challenges, we start by characterizing behaviors of the resource demand, fairness-efficiency trade-off and model deployment, including the dominant resource and recency-aware model reuse. Accordingly, the system incorporates two key modules that work in tandem: an Elastic Multi-Resource Allocator, employs a δ-fairness elastic allocation policy to balance fairness and efficiency by controlling deviations from absolute fairness; and a Model Deployment Manager, updates deployment status by adapting model recency and storage cost. To jointly optimize the actions of both modules, we formulate a mixed-integer nonlinear programming problem. Considering timescale decoupling between the per-slot resource allocation and sporadic model deployment, we decouple it into two subproblems and solve them via meta-heuristics and progressive filling. Extensive experimental results reveal the effectiveness of our system, which improves the utility of resource allocation by up to 17.39%, and the utility of model deployment by 11.22% while ensuring fairness, as well as its scalability under increased user concurrency.
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive Communications and Networking
- Vol. 12, 5063-5076
Networked Edge Resource Orchestration for Mobile AI-Generated Content Services
In this paper, we design an edge-assisted mobile Artificial Intelligence-Generated Content (AIGC) service system. The core of this system is the orchestrated multi-dimensional resource allocation and generative AI model deployment for efficient AIGC service provisioning at the edge. The heterogeneous resource demands of these models strain scarce resources of edge servers. This scarcity calls for a resource-efficient allocation policy whose goal conflicts with ensuring resource fairness for tasks requested by concurrent users. Specifically, such a policy risks enabling resource monopolization by certain tasks, starving others. To address these challenges, we start by characterizing behaviors of the resource demand, fairness-efficiency trade-off and model deployment, including the dominant resource and recency-aware model reuse. Accordingly, the system incorporates two key modules that work in tandem: an Elastic Multi-Resource Allocator, employs a δ-fairness elastic allocation policy to balance fairness and efficiency by controlling deviations from absolute fairness; and a Model Deployment Manager, updates deployment status by adapting model recency and storage cost. To jointly optimize the actions of both modules, we formulate a mixed-integer nonlinear programming problem. Considering timescale decoupling between the per-slot resource allocation and sporadic model deployment, we decouple it into two subproblems and solve them via meta-heuristics and progressive filling. Extensive experimental results reveal the effectiveness of our system, which improves the utility of resource allocation by up to 17.39%, and the utility of model deployment by 11.22% while ensuring fairness, as well as its scalability under increased user concurrency.
- Open Access
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 1833-1844
Leveraging the Power of Conversations: Optimal Key Term Selection in Conversational Contextual Bandits
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 1833-1844
Open AccessLeveraging the Power of Conversations: Optimal Key Term Selection in Conversational Contextual Bandits
- Open Access
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 427-438
A Unified Online-Offline Framework for Co-Branding Campaign Recommendations
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 427-438
Open AccessA Unified Online-Offline Framework for Co-Branding Campaign Recommendations
- Open Access
- 16 June 2025
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 53 (1) , 112-114
Combinatorial Logistic Bandits
Combinatorial multi-armed bandit (CMAB) is a fundamental online learning framework that can optimize cumulative rewards in networked systems under uncertainty. Real-world applications like content delivery and channel allocation often feature binary base arm rewards and nonlinear total reward functions. This paper introduces combinatorial logistic bandits (CLogB), a contextual CMAB framework with the base arm reward modeled as a nonlinear logistic function of the context, and the feedback is governed by a general arm-triggering process. We study CLogB with smooth reward functions, covering applications such as online content delivery, online multi-LLM selection, and dynamic channel allocation. Our first algorithm, CLogUCB, uses a variance-agnostic exploration bonus and achieves a regret bound of Õ( d√(κ KT) ), where d is the feature dimension, κ reflects logistic model nonlinearity, K is the maximum number of triggered arms, and Õ ignores logarithmic factors. This improves on prior results by Õ( √κ ). We further propose VA-CLogUCB, a variance-adaptive enhancement achieving regret bounds of Õ( d√(KT) ) under standard smoothness conditions and Õ( d√T ) under stronger variance conditions, removing dependence on K . For time-invariant feature maps, we enhance computational efficiency by avoiding nonconvex optimization while maintaining Õ(d√T) regret. Experiments on synthetic and real-world datasets validate the superior performance of our algorithms, demonstrating their effectiveness and scalability for real-world networked systems.
- 16 June 2025
- Published by Association for Computing Machinery (ACM) in ACM SIGMETRICS Performance Evaluation Review
- Vol. 53 (1) , 112-114
Open AccessCombinatorial Logistic Bandits
Combinatorial multi-armed bandit (CMAB) is a fundamental online learning framework that can optimize cumulative rewards in networked systems under uncertainty. Real-world applications like content delivery and channel allocation often feature binary base arm rewards and nonlinear total reward functions. This paper introduces combinatorial logistic bandits (CLogB), a contextual CMAB framework with the base arm reward modeled as a nonlinear logistic function of the context, and the feedback is governed by a general arm-triggering process. We study CLogB with smooth reward functions, covering applications such as online content delivery, online multi-LLM selection, and dynamic channel allocation. Our first algorithm, CLogUCB, uses a variance-agnostic exploration bonus and achieves a regret bound of Õ( d√(κ KT) ), where d is the feature dimension, κ reflects logistic model nonlinearity, K is the maximum number of triggered arms, and Õ ignores logarithmic factors. This improves on prior results by Õ( √κ ). We further propose VA-CLogUCB, a variance-adaptive enhancement achieving regret bounds of Õ( d√(KT) ) under standard smoothness conditions and Õ( d√T ) under stronger variance conditions, removing dependence on K . For time-invariant feature maps, we enhance computational efficiency by avoiding nonconvex optimization while maintaining Õ(d√T) regret. Experiments on synthetic and real-world datasets validate the superior performance of our algorithms, demonstrating their effectiveness and scalability for real-world networked systems.