Author name cluster

Jiang Li

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.

13 papers

1 author row

AAAI Conference 2026 Conference Paper

FlorE: Integrating Full Lorentz Group and Directional Offsets for Effective Knowledge Graph Embedding

Zehua Duo
Jiang Li
Xiangdong Su
Guanglai Gao

Knowledge Graph Embedding (KGE) aims to map entities and relationships into a continuous vector space to facilitate reasoning and downstream tasks. Although previous KGE methods based on Euclidean, complex spaces, or hyperbolic spaces have performed well, they still struggle to effectively model Z-Paradox relation patterns which account for a large proportion in each knowledge graph. To address this issue, we propose a novel KGE method **FlorE** which integrates full Lorentz Group and directional offset operation in hyperbolic space for KGE task. Specifically, we incorporates the full Lorentz Group to enable the same relation in knowledge graph (KG) to perform indefinite isometry, thus avoiding the overlapping of entities. Meanwhile, we implement directional offset operation via exponential mapping to transform the relations to the same Lorentz manifold of the entities, thus maintaining geometric consistency for the relations and entities in KG. By integrating these two techniques, FlorE can effectively model the Z-Paradox relation patterns and improve the representation learning ability for KGs. Experiments on the five benchmark datasets demonstrate that our method achieves state-of-the-art performance. For the Z-Paradox relation patterns, the improvement achieves **26.7%**, **15.6%**, **35.4%**, **33.7%**, and **31.5%** on FB15k-237, WN18RR, CoDEx-S, CoDEx-M and CoDEx-L, respectively.

PDF Details DOI

EAAI Journal 2025 Journal Article

A focusing-attention deformable convolution and transformer network with multi-scale contour-render for land cover classification in high-resolution remote-sensing images

Yiming Fan
Dongmei Zhang
Jiang Li
Jianghui Xiao

Land cover classification in High-resolution remote sensing image (HRRSI) aims to achieve pixel-level identification of ground objects on earth. However, the existing methods encounter difficulties in promoting accuracy due to vague contours of HRRSI. This paper proposes a semantic segmentation network called the Focusing-Attention Deformable Convolution and Transformer Network with Multi-Scale Contour-Render (MCFDTNet) to address two main causes of blurred or indistinct contours in HRRSIs. First, we design a Focusing-Attention Module (FAM) in the deformable convolution to enhance the representation of focused regions, mitigating blurring caused by out-of-focus regions and alleviating the foreground-background imbalance. Second, to fully utilize the semantic information in contours, we introduce a novel Double-Branch Multi-Scale Fusion Module (DMFM) to fuse local and global semantic features in the encoder using Multi-Scale Input (MI). We also propose a contour re-rendering mechanism, which dynamically controls the re-rendering region of blurred contour points (CR, ContourRend), aligning the input image edges with the edges in the label image. Comparative experiments were conducted with advanced semantic segmentation models on two datasets: the ISPRS Potsdam dataset and the Gaofen Image Dataset (GID). Ablation studies confirm the effectiveness of the proposed modules. Experimental results demonstrate that MCFDTNet outperforms existing models, achieving improvements of 0. 88 % and 5. 35 % in Mean Intersection over Union (MIoU) on the Potsdam and GID datasets, respectively.

Details DOI

EAAI Journal 2025 Journal Article

Electroencephalogram-based floor vibration serviceability evaluation using convolutional neural network and ensemble learning

Jiang Li
Xuhong Zhou
Jiepeng Liu
Weizhao Tang
Mingyue Xiao
Y. Frank Chen

Discomfort or even panic would occasionally be felt from human-induced vibrations in public buildings such as shopping malls and offices, raising concerns on serviceability. However, the evaluation of human perception is rather subjective. In this study, the electroencephalogram (EEG) was utilized to directly reflect the human perception to floor vibrations. EEG signals and the corresponding perception factors (PFs) were recorded from 12 volunteers experiencing the floor vibrations induced from random heel drop, jump, and hammering. Each EEG time-domain curve was processed with a continuous wavelet transform to combine both time and frequency domain features into one wavelet time-frequency (WTF) image for further analysis. To identify different PFs through WTF images, three convolutional neural networks (CNNs) (Visual geometry group 16-layer network, Google Lenet, and Residual Network 18) and two ensemble learning methods (bagging and stacking) were tested. Based on the comparison of algorithm performances, Google Lenet is recommended to perform the PF recognition as it demonstrates an 89. 4 % accuracy with a running time of 84 s. This study proves the feasibility of using the neuroscientific technique EEG with its time-frequency features to quantitatively grade human perceptions to floor vibration. Furthermore, the correlation establishment between EEG data and perception factors provides a novel approach for evaluating floor vibration serviceability.

Details DOI

JBHI Journal 2025 Journal Article

Optimizing the dynamic administration regimen of prophylactic enoxaparin in critically ill patients using reinforcement learning

Chuanrui Sun
Jiang Li
Zhongheng Zhang
Fengchan Xi
Xiling Wang

This study aims to optimize the dynamic administration regimen of prophylactic enoxaparin in critically ill patients to reduce the risk of VTE, major bleeding, and 30-day all-cause mortality. We developed and internally and externally validated an artificial intelligence (AI) policy utilizing Double dueling deep Q network, using data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database (training and internal test set) and the eICU Collaborative Research Database (eICU-CRD, external test set). We compared the performance among the AI policy, the clinician's policy, the weight-tiered policy, and the fixed 40- mg-once-daily (QD) policy. The rationality and explainability of AI policy were investigated. In the internal test set, the AI policy achieved the highest policy value of 13. 17 and the lowest estimated incidence of all outcomes of 5. 53%, compared with the clinician's policy (9. 81; 23. 60%), the weight-tired policy (12. 14; 10. 72%), and the 40-mg-QD policy (12. 31; 9. 05%). Compared with the clinician's policy, the AI policy was associated with a decreased risk of VTE with statistical significance (OR: 0. 44, 95%CI: 0. 28-0. 69, P<0. 001). The superiority of AI policy was confirmed in the external test set. The SHAP analysis showed that sex, primary diagnosis, time to major surgery, and weight followed by vasopressor administration were the most important features contributing to AI policy. The AI policy could provide effective and clinically reasonable recommendations for the optimal dynamic frequency and dose of prophylactic enoxaparin and could potentially be applied in clinical practice after prospective evaluation.

Details DOI

EAAI Journal 2024 Journal Article

A dual-stream recurrence-attention network with global–local awareness for emotion recognition in textual dialog

Jiang Li
Xiaoping Wang
Zhigang Zeng

In real-world dialog systems, the ability to understand the user’s emotions and interact anthropomorphically is of great significance. Emotion Recognition in Conversation (ERC) is one of the key ways to accomplish this goal and has attracted growing attention. How to model the context in a conversation is a central aspect and a major challenge of ERC tasks. Most existing approaches struggle to adequately incorporate both global and local contextual information, and their network structures are overly sophisticated. For this reason, we propose a simple and effective Dual-stream Recurrence-Attention Network (DualRAN), which is based on Recurrent Neural Network (RNN) and Multi-head ATtention network (MAT). DualRAN eschews the complex components of current methods and focuses on combining recurrence-based methods with attention-based ones. DualRAN is a dual-stream structure mainly consisting of local- and global-aware modules, modeling a conversation simultaneously from distinct perspectives. In addition, we develop two single-stream network variants for DualRAN, i. e. , SingleRANv1 and SingleRANv2. According to the experimental findings, DualRAN boosts the weighted F1 scores by 1. 43% and 0. 64% on the IEMOCAP and MELD datasets, respectively, in comparison to the strongest baseline. On two other datasets (i. e. , EmoryNLP and DailyDialog), our method also attains competitive results.

Details DOI

EAAI Journal 2024 Journal Article

An adaptive spatial–temporal prediction model for landslide displacement based on decomposition architecture

Man Xu
Dongmei Zhang
Jiang Li
Yiping Wu

Landslide displacement forecasting is a core issue in geohazard research, it is particularly challenging for accumulation-type landslides with complex geological patterns. Traditional landslide displacement prediction methods use single-point modeling and often fail to consider the spatial correlation characteristics of each deformation point on the surface of a landslide. On the other hand, they have difficulty in learning the changes caused by rainfall and reservoir water level. To tackle these obstacles, we introduce an adaptive spatial–temporal landslide displacement prediction model based on a decomposition architecture, named Self-Adaptive Unet with Decomposed Temporal Attention Encoder(SAU-DTAE). To effectively separate the features of different scales in time series changes and model them separately, we employ a progressive decomposition architecture based on a Lightweight Temporal Attention Encoder(LTAE). Furthermore, we design a gating mechanism with Sample Entropy (SampEn) to adaptively extract global and local spatial features at multiple scales. By quantifying the spatial complexity, we can achieve adaptive extraction of spatial correlation features. Relevant experiments were conducted with the 2016-2023 Interferometry Synthetic Aperture Radar (InSAR) landslide displacement dataset of the Three Gorges area. The new proposed algorithm was compared and validated against several classical time-series prediction models: Back Propagation(BP) neural network, Long Short Term Memory(LSTM) neural network, Gated Recurrent Unit(GRU), Convolutional LSTM(ConvLSTM), Informer, and Autoformer. The findings from the experiment indicated that our model surpassed the benchmark models, achieving superior prediction results on the test set. The Mean Absolute Error (MAE) was 5. 516 millimeters(mm), the Root Mean Square Error (RMSE) was 3. 856 mm, and the R-Square( R 2 ) was 0. 896.

Details DOI

EAAI Journal 2024 Journal Article

Recognition of local fiber orientation state in prepreg platelet molded composites via deep learning

Richard Larson
Reshad Hoque
Von Jamora
Jiang Li
Sergii G. Kravchenko
Oleksandr G. Kravchenko

This paper presents a novel deep learning approach for reconstruction of local through-the-thickness fiber orientation distribution (FOD) in discontinuous long-fiber, prepreg-platelet molded composites (PPMC). The thermal-residual strains on composite surfaces are used as inputs to train a fully convolutional neural network, named U-Net, to predict spatially varying, local FOD. High fidelity synthetic data was generated via computational simulation of PPMC with stochastic material orientation state and was used for training the deep learning model. The proposed U-Net model allowed for rapid recognition of PPMC morphology by solving the inverse structural mechanics problem of determining the average fiber orientation through the composite thickness based on the provided surface strain measurements. Upon training and validation, the U-Net deep learning model was deployed to rapidly predict complex distributions of the local through-the-thickness FOD in PPMC.

Details DOI

AAAI Conference 2024 Conference Paper

SEER: Backdoor Detection for Vision-Language Models through Searching Target Text and Image Trigger Jointly

Liuwan Zhu
Rui Ning
Jiang Li
Chunsheng Xin
Hongyi Wu

This paper proposes SEER, a novel backdoor detection algorithm for vision-language models, addressing the gap in the literature on multi-modal backdoor detection. While backdoor detection in single-modal models has been well studied, the investigation of such defenses in multi-modal models remains limited. Existing backdoor defense mechanisms cannot be directly applied to multi-modal settings due to their increased complexity and search space explosion. In this paper, we propose to detect backdoors in vision-language models by jointly searching image triggers and malicious target texts in feature space shared by vision and language modalities. Our extensive experiments demonstrate that SEER can achieve over 92% detection rate on backdoor detection in vision-language models in various settings without accessing training data or knowledge of downstream tasks.

PDF Details DOI

EAAI Journal 2023 Journal Article

Randomized block-coordinate adaptive algorithms for nonconvex optimization problems

Yangfan Zhou
Kaizhu Huang
Jiang Li
Cheng Cheng
Xuguang Wang
Amir Hussian
Xin Liu

Nonconvex optimization problems have always been one focus in deep learning, in which many fast adaptive algorithms based on momentum are applied. However, the full gradient computation of high-dimensional feature vector in the above tasks become prohibitive. To reduce the computation cost for optimizers on nonconvex optimization problems typically seen in deep learning, this work proposes a randomized block-coordinate adaptive optimization algorithm, named RAda, which randomly picks a block from the full coordinates of the parameter vector and then sparsely computes its gradient. We prove that RAda converges to a δ -accurate solution with the stochastic first-order complexity of O ( 1 / δ 2 ), where δ is the upper bound of the gradient’s square, under nonconvex cases. Experiments on public datasets including CIFAR-10, CIFAR-100, and Penn TreeBank, verify that RAda outperforms the other compared algorithms in terms of the computational cost.

Details DOI

JBHI Journal 2022 Journal Article

Building a Risk Prediction Model for Postoperative Pulmonary Vein Obstruction via Quantitative Analysis of CTA Images

Yuchen Pei
Guocheng Shi
Wenjin Xia
Chen Wen
Dazhen Sun
Fang Zhu
Jiang Li
Zhongqun Zhu

Total anomalous pulmonary venous connection (TAPVC) is a rare but mortal congenital heart disease in children and can be repaired by surgical operations. However, some patients may suffer from pulmonary venous obstruction (PVO) after surgery with insufficient blood supply, necessitating special follow-up strategy and treatment. Therefore, it is a clinically important yet challenging problem to predict such patients before surgery. In this paper, we address this issue and propose a computational framework to determine the risk factors for postoperative PVO (PPVO) from computed tomography angiography (CTA) images and build the PPVO risk prediction model. From clinical experiences, such risk factors are likely from the left atrium (LA) and pulmonary vein (PV) of the patient. Thus, 3D models of LA and PV are first reconstructed from low-dose CTA images. Then, a feature pool is built by computing different morphological features from 3D models of LA and PV, and the coupling spatial features of LA and PV. Finally, four risk factors are identified from the feature pool using the machine learning techniques, followed by a risk prediction model. As a result, not only PPVO patients can be effectively predicted but also qualitative risk factors reported in the literature can now be quantified. Finally, the risk prediction model is evaluated on two independent clinical datasets from two hospitals. The model can achieve the AUC values of 0. 88 and 0. 87 respectively, demonstrating its effectiveness in risk prediction.

Details DOI

AAAI Conference 2022 Conference Paper

Hibernated Backdoor: A Mutual Information Empowered Backdoor Attack to Deep Neural Networks

Rui Ning
Jiang Li
Chunsheng Xin
Hongyi Wu
Chonggang Wang

We report a new neural backdoor attack, named Hibernated Backdoor, which is stealthy, aggressive and devastating. The backdoor is planted in a hibernated mode to avoid being detected. Once deployed and fine-tuned on end-devices, the hibernated backdoor turns into the active state that can be exploited by the attacker. To the best of our knowledge, this is the first hibernated neural backdoor attack. It is achieved by maximizing the mutual information (MI) between the gradients of regular and malicious data on the model. We introduce a practical algorithm to achieve MI maximization to effectively plant the hibernated backdoor. To evade adaptive defenses, we further develop a targeted hibernated backdoor, which can only be activated by specific data samples and thus achieves a higher degree of stealthiness. We show the hibernated backdoor is robust and cannot be removed by existing backdoor removal schemes. It has been fully tested on four datasets with two neural network architectures, compared to five existing backdoor attacks, and evaluated using seven backdoor detection schemes. The experiments demonstrate the effectiveness of the hibernated backdoor attack under various settings.

PDF Details

JBHI Journal 2015 Journal Article

A Robust Deep Model for Improved Classification of AD/MCI Patients

Feng Li
Loc Tran
Kim-Han Thung
Shuiwang Ji
Dinggang Shen
Jiang Li

Accurate classification of Alzheimer's disease (AD) and its prodromal stage, mild cognitive impairment (MCI), plays a critical role in possibly preventing progression of memory impairment and improving quality of life for AD patients. Among many research tasks, it is of a particular interest to identify noninvasive imaging biomarkers for AD diagnosis. In this paper, we present a robust deep learning system to identify different progression stages of AD patients based on MRI and PET scans. We utilized the dropout technique to improve classical deep learning by preventing its weight coadaptation, which is a typical cause of overfitting in deep learning. In addition, we incorporated stability selection, an adaptive learning factor, and a multitask learning strategy into the deep learning framework. We applied the proposed method to the ADNI dataset, and conducted experiments for AD and MCI conversion diagnosis. Experimental results showed that the dropout technique is very effective in AD diagnosis, improving the classification accuracies by 5. 9% on average as compared to the classical deep learning methods.

Details DOI

YNIMG Journal 2011 Journal Article

In vivo retinotopic mapping of superior colliculus using manganese-enhanced magnetic resonance imaging

Kevin C. Chan
Jiang Li
Phillis Kau
Iris Y. Zhou
Matthew M. Cheung
Condon Lau
Jian Yang
Kwok-fai So

The superior colliculus (SC) is a dome-shaped subcortical laminar structure in the mammalian midbrain, whose superficial layers receive visual information from the retina in a topological order. Despite the increasing number of studies investigating retinotopic projection in visual brain development and disorders, in vivo, high-resolution 3D mapping of topographic organization in the subcortical visual nuclei has not yet been available. This study explores the capability of 3D manganese-enhanced MRI (MEMRI) at 200μm isotropic resolution for in vivo retinotopic mapping of the rat SC upon partial transection of the intraorbital optic nerve. One day after intravitreal Mn2+ injection into both eyes, animals with partial transection at the right superior intraorbital optic nerve in Group 1 (n=8) exhibited a significantly lower T1-weighted signal intensity in the lateral region of the left SC compared to the left medial SC and right control SC. Partial transection toward the temporal or nasal region of the right intraorbital optic nerve in Group 2 (n=7) led to T1-weighted hypointensity in the rostral or caudal region of the left SC, whereas a clear border was observed separating 2 halves of the left SC in all groups. Previous histological and electrophysiological studies showed that the retinal ganglion cell axons emanating from superior, inferior, nasal and temporal retina projected respectively to the contralateral lateral, medial, caudal and rostral SC in rodents. While this topological pattern is preserved in the intraorbital optic nerve, it was shown that partial transection of the superior intraorbital optic nerve led to primary injury predominantly in the superior but not inferior retina and optic nerve. The results of this study demonstrated the sensitivity of submillimeter-resolution MEMRI for in vivo, 3D mapping of the precise retinotopic projections in SC upon reduced anterograde axonal transport of Mn2+ ions from localized regions of the anterior visual pathways to the subcortical midbrain nuclei. Future MEMRI studies are envisioned that measure the topographic changes in brain development, diseases, plasticity and regeneration therapies in a global and longitudinal setting.

Details DOI