A5100397027- Face and Expression Recognition
- Text and Document Classification Technologies
- Anomaly Detection Techniques and Applications
- Privacy-Preserving Technologies in Data
- Advanced Clustering Algorithms Research
- Advanced Algorithms and Applications
- Bayesian Modeling and Causal Inference
- Stochastic Gradient Optimization Techniques
- Machine Learning and Data Classification
- Time Series Analysis and Forecasting
- Cryptography and Data Security
- Imbalanced Data Classification Techniques
- Gene expression and cancer classification
- Network Security and Intrusion Detection
- Advanced Image and Video Retrieval Techniques
- Data Mining Algorithms and Applications
- Domain Adaptation and Few-Shot Learning
- Advanced Graph Neural Networks
- Neural Networks and Applications
- Bioinformatics and Genomic Networks
- Complex Network Analysis Techniques
- Mobile Crowdsensing and Crowdsourcing
- Reinforcement Learning in Robotics
- Rough Sets and Fuzzy Logic
- Image Retrieval and Classification Techniques
Guangdong University of Technology
2013-2025
Shantou University
2021-2025
Xi'an University of Technology
2019
South China University of Technology
2003-2016
Foshan University
2016
Nanjing University
2014
Research on social bot detection plays a crucial role in maintaining the order and reliability of information dissemination while increasing trust interactions. The current mainstream models rely black-box neural network technology, e.g., Graph Neural Network, Transformer, etc., which lacks interpretability. In this work, we present UnDBot, novel unsupervised, interpretable, yet effective practical framework for detecting bots. This is built upon structural theory. We begin by designing...
In recent years, incomplete multi-view clustering (IMVC), which studies the challenging problem on missing views, has received growing research interests. Previous IMVC methods suffer from following issues: (1) inaccurate imputation for data, leads to suboptimal performance, and (2) most existing models merely consider explicit presence of graph structure in ignoring fact that latent graphs different views also provide valuable information task. To overcome such challenges, we present a...
Differential privacy is a promising privacy-preserving paradigm for statistical query processing over sensitive data. It works by injecting random noise into each result, such that it provably hard the adversary to infer presence or absence of any individual record from published noisy results. The main objective in differentially private maximize accuracy results, while satisfying guarantees. Previous work, notably matrix mechanism [16], has suggested batch correlated queries as whole can...
Covalent adaptable networks incorporating supramolecular interactions are constructed to produce poly(thioctic acid)/aluminum (poly(TA)/Al) self-healing and stretchable composite elastomers for thermal management of advanced electronic devices.
Drug combinations play very important roles in cancer therapy, as they can enhance curative efficacy and overcome drug resistance. Due to the increasing size of combinatorial space, experimental screening for all becomes infeasible practice. Therefore, there is a great need develop accurate computational approaches that predict potential direct screening. In this paper, we propose novel method called GNNSynergy learn embeddings synergy prediction. Given specific cell line, multi-view graph...
The task of outlier detection is to identify data objects that are markedly different from or inconsistent with the normal set data. Most existing solutions typically build a model using and outliers do not fit represented very well. However, in addition data, there also exist limited negative examples many applications, may be corrupted such imperfectly labeled. These make far more difficult than traditional ones. This paper presents novel approach address imperfect labels incorporate...
In the era of big data, we can easily access information from multiple views which may be obtained different sources or feature subsets. Generally, provide complementary for learning tasks. Thus, multi-view facilitate process and is prevalent in a wide range application domains. For example, medical science, measurements series examinations are documented each subject, including clinical, imaging, immunologic, serologic cognitive measures sources. Specifically, brain diagnosis, have...
Support vector machines (SVMs), which were originally designed for binary classifications, are an excellent tool machine learning. For the multiclass they usually converted into ones before can be used to classify examples. In one-against-one algorithm with SVMs, there exists unclassifiable region where data samples cannot classified by its decision function. This paper extends handle this problem. We also give convergence and computational complexity analysis of proposed method. Finally,...
Multiple-instance learning (MIL) is a generalization of supervised that attempts to learn useful information from bags instances. In MIL, the true labels instances in positive are not available for training. This leads critical challenge, namely, handling which ambiguous (ambiguous instances). To deal with these instances, we propose novel MIL approach, called similarity-based multiple-instance (SMILE). Instead eliminating number training classifier, as done some previous works, SMILE...
Data mining technology can be used to dig out potential and valuable information from massive data, support vector machine (SVM) is one of the most widely efficient methods in field data classification.However, training set often contains sensitive attributes, traditional method SVM reveals individual privacy information.In view low prediction accuracy poor versatility existing classifiers with protection, this paper proposed a new for differential protection.The algorithm first solved dual...
This paper presents a novel framework to uncertain one-class learning and concept summarization on data streams. Our proposed consists of two parts. First, we put forward cope with uncertainty. We first propose local kernel-density-based method generate bound score for each instance, which refines the location corresponding then construct an classifier (UOCC) by incorporating generated into SVM-based phase. Second, support vectors (SVs)-based clustering technique summarize user from history...
Multiview graph clustering (MGC) methods are increasingly being studied due to the explosion of multiview data with structural information. The critical point MGC is better utilize view-specific and view-common information in features graphs multiple views. However, existing works have an inherent limitation that they unable concurrently consensus across feature To address this issue, we propose a variational generator for (VGMGC). Specifically, novel proposed extract common among graphs....
Mining temporal patterns from user behaviors has long been investigated, but most of the existing work centers on single-type user-item interactions, such as purchase or click, which fails to take advantage user’s diversified interests revealed by various types behavior. However, capturing different behavior sequences and modeling complex inter-correlation between them are non-trivial tasks, high sparsity type-related multi-seasonality individual time-variant dependency multi-type activities...
SVDD has been proved a powerful tool for outlier detection. However, in detecting outliers on multi-distribution data, namely there are distinctive distributions the it is very challenging to generate hyper-sphere distinguishing from normal data. Even if such can be identified, its performance usually not good enough. This paper proposes an multi-sphere approach, named MS-SVDD, detection First, adaptive sphere method proposed detect data dataset. The partitioned terms of identified...
Differential privacy is a promising privacy-preserving paradigm for statistical query processing over sensitive data. It works by injecting random noise into each result such that it provably hard the adversary to infer presence or absence of any individual record from published noisy results. The main objective in differentially private maximize accuracy results while satisfying guarantees. Previous work, notably Li et al. [2010], has suggested that, with an appropriate strategy, batch...
Depression is a common mental disorder that seriously affects patients' social function and daily life. Its accurate diagnosis remains big challenge in depression treatment. In this study, we used electroencephalography (EEG) functional near-infrared spectroscopy (fNIRS) measured the whole brain EEG signals forehead hemodynamic from 25 patients 30 healthy subjects during resting state. On one hand, explored network properties, found clustering coefficient local efficiency of delta theta...