Institution profile

Queen Mary University of London

Academic institutioneurope · gb
Official website
Research library697linked papers
Opportunities0open roles
Selected work

Representative Papers

OmniBench: Towards The Future of Universal Omni-Language Models

Sep 23, 2024arXiv.org

Existing open-source multimodal large language models (MLLMs) exhibit significant deficiencies in joint visual-auditory-textual understanding and reasoning, achieving only ~50% instruction-following accuracy on trilingual multimodal tasks. Method: We introduce OmniBench—the first benchmark for trilingual multimodal collaborative reasoning—and formalize the omni-language model (OLM), a unified architecture capable of jointly processing visual, auditory, and textual (V-A-T) inputs. We construct OmniBench via expert human annotation across diverse trilingual multimodal tasks and curate OmniInstruct, a large-scale instruction-tuning dataset comprising 96K samples. Our methodology integrates cross-modal alignment modeling, trilingual multimodal instruction tuning, and a human-in-the-loop evaluation framework. Contribution/Results: Experiments reveal severe generalization limitations of current open-source OLMs on trilingual multimodal tasks; OmniInstruct substantially improves their reasoning performance. This work establishes a novel evaluation paradigm, provides high-quality resources, and outlines a technical pathway for advancing trilingual multimodal foundation models.

9 citations2 influentialRead paper

When xURLLC Meets NOMA: A Stochastic Network Calculus Perspective

Jun 01, 2024IEEE Communications Magazine

To address the stringent requirements of ultra-low latency, ultra-high reliability, and fresh information (quantified by Age of Information, AoI) in xURLLC systems, this paper proposes a NOMA-assisted uplink architecture. It introduces stochastic network calculus (SNC) into the NOMA-xURLLC domain for the first time, establishing a unified theoretical framework that enables joint statistical QoS provisioning (SQP) for latency, AoI, and reliability tail distributions. Furthermore, we propose an SQP-driven power optimization paradigm, leveraging convex optimization and a customized power allocation algorithm to minimize uplink transmit power while satisfying multi-dimensional QoS constraints. Simulation results demonstrate that the proposed scheme outperforms conventional orthogonal multiple access across all key metrics—latency, AoI, reliability, and energy efficiency.

7 citationsRead paper

Should the Timing of Inspections be Predictable?

Apr 03, 2023ACM Conference on Economics and Computation

This paper examines how a principal should design the timing of monitoring inspections to incentivize agent effort—specifically, whether inspections should be deterministic or stochastic. Method: Building on principal–agent theory and dynamic game modeling, we incorporate incentive compatibility constraints and optimal contract design, and—novelly—classify tasks exogenously by intrinsic nature into “breakthrough-oriented” (e.g., innovation) and “failure-avoidance–oriented” (e.g., risk control). Contribution/Results: We prove that deterministic inspections dominate for breakthrough-oriented tasks, as they reinforce agents’ long-term effort expectations; conversely, stochastic inspections are optimal for failure-avoidance–oriented tasks, as they mitigate strategic window-avoidance behavior. This reveals a structural alignment mechanism between inspection predictability and task type, providing a rigorous theoretical foundation for differentiated regulatory policy design.

6 citations1 influentialRead paper

Impartial Games: A Challenge for Reinforcement Learning

May 25, 2022arXiv.org

This work identifies the fundamental cause of generalization failure in AlphaZero-style reinforcement learning on impartial games (e.g., Nim): neural networks relying on local observations cannot implicitly learn global, non-local abstract functions—such as parity—that determine game-theoretic outcomes, and state-outcome correlations vanish identically. We construct an AlphaZero variant integrating self-play training, Monte Carlo Tree Search, and residual CNNs, augmented with state-masking analysis and systematic generalization diagnostics. For the first time, we rigorously establish that the decoupling between local observability and global win-loss determination constitutes the core bottleneck for RL success in such domains, while also linking this failure to data skew and label noise. Experiments show convergence on small Nim instances, but training efficiency collapses catastrophically with scale; moreover, value networks fail to infer outcomes from partial states, exhibiting markedly lower robustness than in biased games like Chess or Go.

6 citations1 influentialRead paper

A Probabilistic Digital Twin of UK En Route Airspace for Training and Evaluating AI Agents for Air Traffic Control

Jan 06, 2026AIAA SCITECH 2026 Forum

This study addresses the critical gap in current AI research for air traffic control—namely, the absence of a safe, high-fidelity, and quantifiable virtual environment for training and evaluation. The authors present the first probabilistic digital twin system tailored to UK en-route airspace, integrating historical and real-time operational data with physics-informed machine learning models to faithfully reproduce realistic traffic scenarios and enable human–AI collaborative assessment. Innovatively, the framework incorporates a structured validation methodology grounded in trustworthiness and ethical safeguards, delivering a unified, high-speed, standardized testing platform capable of simulating up to 200× real-time speed. Through a Python Gym interface, an interactive human-in-the-loop interface, and quantitative performance metrics, the system facilitates rapid iteration of AI agents and controller-led capability evaluation in a high-fidelity airspace, laying the groundwork for advanced automation in air traffic management.

6 citationsRead paper
Recent publications

Latest Papers