A5100398281- Metaheuristic Optimization Algorithms Research
- Topic Modeling
- Advanced Multi-Objective Optimization Algorithms
- Multimodal Machine Learning Applications
- Data Management and Algorithms
- Evolutionary Algorithms and Applications
- Advanced Graph Neural Networks
- Advanced Text Analysis Techniques
- Advanced Image and Video Retrieval Techniques
- Natural Language Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Recommender Systems and Techniques
- Web Data Mining and Analysis
- Geographic Information Systems Studies
- Human Mobility and Location-Based Analysis
- Text and Document Classification Technologies
- Nutrition and Health in Aging
- Intelligent Tutoring Systems and Adaptive Learning
- Privacy-Preserving Technologies in Data
- Ubiquitin and proteasome pathways
- Advanced Computational Techniques and Applications
- Traffic Prediction and Management Techniques
- Peer-to-Peer Network Technologies
- Explainable Artificial Intelligence (XAI)
- Human Pose and Action Recognition
East China Normal University
2016-2025
University of Science and Technology of China
2014-2025
Lanzhou University Second Hospital
2022-2025
Army Medical University
2022-2025
Lanzhou University
2022-2025
Xinqiao Hospital
2025
Daping Hospital
2022-2024
Tianjin University of Technology
2023-2024
Harbin Institute of Technology
2022-2024
First Affiliated Hospital of Fujian Medical University
2022-2024
Scene graph generation (SGG) aims to detect objects in an image along with their pairwise relationships. There are three key properties of scene that have been underexplored recent works: namely, the edge direction information, difference priority between nodes, and long-tailed distribution Accordingly, this paper, we propose a Graph Property Sensing Network (GPS-Net) fully explores these for SGG. First, novel message passing module augments node feature node-specific contextual information...
Personalized Route Recommendation (PRR) aims to generate user-specific route suggestions in response users' queries. Early studies cast the PRR task as a pathfinding problem on graphs, and adopt adapted search algorithms by integrating heuristic strategies. Although these methods are effective some extent, they require setting cost functions with heuristics. In addition, it is difficult utilize useful context information procedure. To address issues, we propose using neural networks...
Multimodal optimization problems (MMOPs) are common in real-world applications and involve identifying multiple optimal solutions for decision makers to choose from. The core requirement dealing with such is balance the ability of exploration global space exploitation areas. In this paper, based on differential evolution (DE), we propose a novel algorithm focusing formulation, balance, keypoint species MMOPs, called FBK-DE. First, nearest-better clustering (NBC) used divide population into...
Visual grounding focuses on establishing fine-grained alignment between vision and natural language, which has essential applications in multimodal reasoning systems. Existing methods use pre-trained query-agnostic visual backbones to extract feature maps independently without considering the query information. We argue that features extracted from really needed for are inconsistent. One reason is there differences pre-training tasks grounding. Moreover, since query-agnostic, it difficult...
Automatically solving math word problems is a crucial task for exploring the intelligence levels of machines in general AI domain. It highly challenging since it requires not only natural language understanding but also mathematical expression inference. Existing solutions usually explore sequence-to-sequence models to generate expressions, where are simply encoded sequentially. However, such generally far from enough as similar humans and lead incorrect answers. To this end, paper, we...
Scene graph generation (SGG) aims to detect objects and predict their pairwise relationships within an image. Current SGG methods typically utilize neural net-works (GNNs) acquire context information between ob-jects/relationships. Despite effectiveness, however, current only assume scene homophily while ignoring heterophily. Accordingly, in this paper, we propose a novel Heterophily Learning Network (HL-Net) comprehensively explore the heterophily be-tween objects/relationships graphs. More...
Scene graph generation (SGG) aims to detect objects and predict the relationships between each pair of objects. Existing SGG methods usually suffer from several issues, including 1) ambiguous object representations, as neural network-based message passing (GMP) modules are typically sensitive spurious inter-node correlations, 2) low diversity in relationship predictions due severe class imbalance a large number missing annotations. To address both problems, this paper, we propose...
Cognitive diagnosis (CD) aims to reveal the proficiency of students on specific knowledge concepts and traits test exercises (e.g., difficulty). It plays a critical role in intelligent education systems by supporting personalized learning guidance. However, recent developments CD mostly concentrate improving accuracy diagnostic results often overlook important practical task: domain-level zero-shot cognitive (DZCD). The primary challenge DZCD is deficiency student behavior data target domain...
The concept of possible sarcopenia (PS) was recently introduced to enable timely intervention in settings without the technologies required make a full diagnosis sarcopenia. This study aimed investigate association between PS and all-cause mortality patients with solid cancer. Retrospective observational study. 13,736 16 types cancer who were ≥18 years old. presence both low calf circumference (men <34 cm or women <33 cm) handgrip strength <28 kg <18 kg) considered indicate PS. Harrell's...
Multimodal optimization problems (MMOPs) are common with multiple optimal solutions. In this article, a novel method of population division, called nearest-better-neighbor clustering (NBNC), is proposed, which can reduce the risk more than one species locating same peak. The key idea NBNC to construct raw by linking each individual better within neighborhood, and final formulated merging dominated species. Furthermore, algorithm proposed NBNC-PSO-ES, combines advantages exploration in...
Building a good feature space is essential for the metric-based few-shot algorithms to recognize novel class with only few samples. The often built by Convolutional Neural Networks (CNNs). However, CNNs primarily focus on local information limited receptive field, and global generated distant pixels not well used. Meanwhile, having understanding of current task focusing distinct regions same sample different queries are important classification. To tackle these problems, we propose Cross...
In this paper, we extend our previous work on the power series method for computing backstepping kernels. Our first contribution is development of initial steps towards a MATLAB toolbox dedicated to kernel computation. This would exploit MATLAB's linear algebra and sparse matrix manipulation features enhanced efficiency; findings show considerable improvements in computational speed with respect use symbolical software without loss precision at high orders. Additionally, tackle limitations...
We aimed to investigate the role of Rnf40 in hypertension-induced cerebrovascular endothelial barrier dysfunction and cognitive impairment. employed microarray data analysis integrated bioinformatics databases identify a novel E3 ligase, Rnf40, that targets Parkin. To understand RNF40 cell damage, we used pAAV-hFLT1-MCS-EGFP-3×Flag-mir30shRnf40 establish an Rnf40-deficient model spontaneously hypertensive rats (SHRs). also evaluated function, cerebral blood flow, performance. observed...
Tropical wetlands account for &#8764;20% of the global total methane (CH4) emissions, but uncertainties remain in emission estimation due to inaccurate representation wetland spatiotemporal variations. Based on latest satellite observational inundation data, we constructed a model map long-term time series extents over Sudd floodplain, which has recently been identified as an important source CH4 emissions. Our analysis reveals annual, extent 5.73 &#177; 2.05 &#215; 104 km2...
Skyline query processing for location-based services, which considers both spatial and nonspatial attributes of the objects being queried, has recently received increasing attention. Existing solutions focus on solving point- or line-based skyline queries, in location is an exact point a line segment. However, due to privacy concerns limited precision localization devices, input user often range. This paper studies new problem how process such range-based queries. Two novel algorithms are...
Web objects, often associated with descriptive text documents, are increasingly being geo-tagged. A spatial keyword top-k query retrieves the best k such objects according to a scoring function that considers both distance and textual similarity. However, it is in some cases difficult for users identify exact keywords describe their intent. After user issues an initial gets back result, may find expected missing wonder why. Answering resulting why-not questions can aid retrieving better...
Learning informative representations for educational questions is a fundamental problem in online learning systems, which can promote many applications, e.g., difficulty estimation. Most solutions integrate all information of one question together following supervised manner, where the representation results are unsatisfactory sometimes due to issues. First, they cannot ensure presentation ability scarcity labeled data. Then, label-dependent have poor feasibility be transferred. Moreover,...
Dynamic and multimodal features are two important properties widely existed in many real-world optimization problems. The former illustrates that the objectives and/or constraints of problems change over time, while latter means there is more than one optimal solution (sometimes including accepted local solutions) each environment. dynamic (DMMOPs) have both these characteristics, which been studied field evolutionary computation swarm intelligence for years, attract attention. Solving such...