Author name cluster

Fu Xiao

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

5 papers

1 author row

EAAI Journal 2026 Journal Article

Leveraging large language model agents for cost-effective sensor data handling and urban traffic navigation

Jie Deng
Yongxu Zhu
Fu Xiao

Large language models have recently demonstrated strong reasoning capabilities beyond natural language processing, creating new opportunities for intelligent handling of heterogeneous data in engineering systems. This work studies the application of artificial intelligence, specifically large language model? based agents, to sensor data interpretation and urban traffic navigation. This paper presents a unified agent-based platform that leverages large language model reasoning to interpret sensor data and support urban traffic navigation. The work introduces two complementary contributions. First, a domain-adaptive prompting mechanism is proposed to enable large language models to reason over diverse sensor modalities by transforming numerical and symbolic measurements into structured decision contexts. This approach allows the model to generalize across previously unseen sensor types while maintaining robust performance. Second, a modular agent-oriented system architecture is designed to integrate large language model reasoning with traffic control functions and Internet of Things middleware. The architecture supports cost-effective deployment, scalable coordination, and real-time responsiveness in both centralized and distributed environments. Empirical evaluations demonstrate three key outcomes. The proposed platform achieves one hundred percent accuracy in interpreting five representative sensor data types within thirty seconds. For urban routing tasks, it attains up to ninety-two percent accuracy using pretrained models without fine-tuning, with response times ranging from twenty-three to one hundred and three seconds depending on task complexity. In simulated traffic networks, the system reduces travel time by up to eighteen percent through adaptive route recommendations. These results indicate that combining domain-adaptive prompting with an agent-based architecture provides a practical and scalable alternative to traditional software-intensive approaches for smart city traffic management.

Details DOI

AAAI Conference 2026 Conference Paper

Transferable Backdoor Attacks for Code Models via Sharpness-Aware Adversarial Perturbation

Shuyu Chang
Haiping Huang
Yanjun Zhang
Yujin Huang
Fu Xiao
Leo Yu Zhang

Code models are increasingly adopted in software development but remain vulnerable to backdoor attacks via poisoned training data. Existing backdoor attacks on code models face a fundamental trade-off between transferability and stealthiness. Static trigger-based attacks insert fixed dead code patterns that transfer well across models and datasets but are easily detected by code-specific defenses. In contrast, dynamic trigger-based attacks adaptively generate context-aware triggers to evade detection but suffer from poor cross-dataset transferability. Moreover, they rely on unrealistic assumptions of identical data distributions between poisoned and victim training data, limiting their practicality. To overcome these limitations, we propose Sharpness-aware Transferable Adversarial Backdoor (STAB), a novel attack that achieves both transferability and stealthiness without requiring complete victim data. STAB is motivated by the observation that adversarial perturbations in flat regions of the loss landscape transfer more effectively across datasets than those in sharp minima. To this end, we train a surrogate model using Sharpness-Aware Minimization to guide model parameters toward flat loss regions, and employ Gumbel-Softmax optimization to enable differentiable search over discrete trigger tokens for generating context-aware adversarial triggers. Experiments across three datasets and two code models show that STAB outperforms prior attacks in terms of transferability and stealthiness. It achieves a 73.2% average attack success rate after defense, outperforming static trigger–based attacks that fail under defense. STAB also surpasses the best dynamic trigger–based attack by 12.4% in cross-dataset attack success rate and maintains performance on clean inputs.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

Learning from Disjoint Views: A Contrastive Prototype Matching Network for Fully Incomplete Multi-View Clustering

Yiming Wang
Qun Li
Dongxia Chang
Jie Wen
Hua Dai
Fu Xiao
Yao Zhao

Multi-view clustering aims to enhance clustering performance by leveraging information from diverse sources. However, its practical application is often hindered by a barrier: the lack of correspondences across views. This paper focuses on the understudied problem of fully incomplete multi-view clustering (FIMC), a scenario where existing methods fail due to their reliance on partial alignment. To address this problem, we introduce the Contrastive Prototype Matching Network (CPMN), a novel framework that establishes a new paradigm for cross-view alignment based on matching high-level categorical structures. Instead of aligning individual instances, CPMN performs a more robust cluster prototype alignment. CPMN first employs a correspondence-free graph contrastive learning approach, leveraging mutual $k$-nearest neighbors (MNN) to uncover intrinsic data structures and establish initial prototypes from entirely unpaired views. Building on the prototypes, we introduce a cross-view prototype graph matching stage to resolve category misalignment and forge a unified clustering structure. Finally, guided by this alignment, we devise a prototype-aware contrastive learning mechanism to promote semantic consistency, replacing the reliance on the initial MNN-based structural similarity. Extensive experiments on benchmark datasets demonstrate that our method significantly outperforms various baselines and ablation variants, validating its effectiveness.

PDF Details

EAAI Journal 2024 Journal Article

Forecasting backdraft with multimodal method: Fusion of fire image and sensor data

Tianhang Zhang
Fangqiang Ding
Zilong Wang
Fu Xiao
Chris Xiaoxuan Lu
Xinyan Huang

Experienced firefighters can fuse the flame image, smoke pattern, and varying temperature, sound, and odour in complex and fast-changing fire scenes to foresee flashover and explosion. This study mimics firefighters and proposes a novel transformer algorithm for the fusion of fire images and temperature sensor data to forecast the backdraft explosion in a building fire. The model of backdraft forecast is demonstrated with full-scale fire tests. After training 2674 fire scenarios with various fire intensities and images from various view angles, the Fusion-Transformer model can forecast the risk of backdraft with an overall accuracy of 84%. Moreover, the occurrence time and explosion scale of backdraft can be predicted with the Mean Absolute Error (MAE) of 1. 6 s and 0. 14 m, respectively. Compared with the single modal model, the fusion of fire images and temperature sensor data improves the accuracy of backdraft forecast by over 50%. This work demonstrates the use of a transformer algorithm in forecasting fire evolution and critical events. It also bridges the gap between data fusion methods and fire forecast, which inspires future universal AI-driven smart firefighting practices.

Details DOI

IJCAI Conference 2022 Conference Paper

Dite-HRNet: Dynamic Lightweight High-Resolution Network for Human Pose Estimation

Qun Li
Ziyi Zhang
Fu Xiao
Feng Zhang
Bir Bhanu

A high-resolution network exhibits remarkable capability in extracting multi-scale features for human pose estimation, but fails to capture long-range interactions between joints and has high computational complexity. To address these problems, we present a Dynamic lightweight High-Resolution Network (Dite-HRNet), which can efficiently extract multi-scale contextual information and model long-range spatial dependency for human pose estimation. Specifically, we propose two methods, dynamic split convolution and adaptive context modeling, and embed them into two novel lightweight blocks, which are named dynamic multi-scale context block and dynamic global context block. These two blocks, as the basic component units of our Dite-HRNet, are specially designed for the high-resolution networks to make full use of the parallel multi-resolution architecture. Experimental results show that the proposed network achieves superior performance on both COCO and MPII human pose estimation datasets, surpassing the state-of-the-art lightweight networks. Code is available at: https: //github. com/ZiyiZhang27/Dite-HRNet.

PDF Details DOI