A5100752256- Advanced Text Analysis Techniques
- Smart Grid Security and Resilience
- Adversarial Robustness in Machine Learning
- Sentiment Analysis and Opinion Mining
- Software Testing and Debugging Techniques
- Physical Unclonable Functions (PUFs) and Hardware Security
- Advanced Image and Video Retrieval Techniques
- Domain Adaptation and Few-Shot Learning
- Face and Expression Recognition
- Image Retrieval and Classification Techniques
- Advanced Malware Detection Techniques
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Cell Image Analysis Techniques
- Network Security and Intrusion Detection
- Online Learning and Analytics
- Software Reliability and Analysis Research
- Advanced Algorithms and Applications
- Advanced machining processes and optimization
- AI in cancer detection
- Gene expression and cancer classification
- Advanced Sensor and Energy Harvesting Materials
- Computational and Text Analysis Methods
- Text and Document Classification Technologies
- Cryptography and Data Security
ShanghaiTech University
2023-2025
Southern Medical University
2024
Wuhan University
2019-2024
Southeast University
2024
Zhongnan Hospital of Wuhan University
2024
Fudan University
2023-2024
Shantou University
2024
Tianjin University
2024
China University of Geosciences
2022-2023
Zhejiang Normal University
2023
In this paper, we propose and evaluate the application of unsupervised machine learning to anomaly detection for a Cyber-Physical System (CPS). We compare two methods: Deep Neural Networks (DNN) adapted time series data generated by CPS, one-class Support Vector Machines (SVM). These methods are evaluated against from Secure Water Treatment (SWaT) testbed, scaled-down but fully operational raw water purification plant. For both methods, first train detectors using log SWaT operating under...
Cyber-physical systems (CPS) consist of sensors, actuators, and controllers all communicating over a network; if any subset becomes compromised, an attacker could cause significant damage. With access to data logs model the CPS, physical effects attack potentially be detected before damage is done. Manually building that accurate enough in practice, however, extremely difficult. In this paper, we propose novel approach for constructing models CPS automatically, by applying supervised machine...
Aspect Sentiment Triplet Extraction (ASTE) is a new fine-grained sentiment analysis task that aims to extract triplets of aspect terms, sentiments, and opinion terms from review sentences. Recently, span-level models achieve gratifying results on ASTE by taking advantage the predictions all possible spans. Since spans significantly increases number potential candidates, it crucial challenging efficiently triplet elements among them. In this paper, we present bidirectional network which...
GPS trajectories are the essential foundations for many trajectory-based applications. Most applications require a large number of high sample rate to achieve good performance. However, real-life collected with low due energy concern or other constraints. We study task trajectory recovery in this paper as means increase trajectories. existing works on follow sequence-to-sequence diagram, an encoder encode and decoder recover real points trajectory. these ignore topology road network only use...
Abstract The floor constitutes one of the largest areas within a building with which users interact most frequently in daily activities. Employing sensors is vital for smart‐building digital twins, wherein triboelectric nanogenerators demonstrate wide application potential due to their good performance and self‐powering characteristics. However, sensing stability, reliability, multimodality require further enhancement meet rapidly evolving demands. Thus, this work introduces multimodal...
Abstract Benefiting from the widespread potential applications in era of Internet Thing and metaverse, triboelectric piezoelectric nanogenerators (TENG & PENG) have attracted considerably increasing attention. Their outstanding characteristics, such as self-powered ability, high output performance, integration compatibility, cost-effectiveness, simple configurations, versatile operation modes, could effectively expand lifetime vastly distributed wearable, implantable, environmental...
Self-supervised learning has emerged as a highly effective approach in the fields of natural language processing and computer vision. It is also applicable to brain signals such electroencephalography (EEG) data, given abundance available unlabeled data that exist wide spectrum real-world medical applications ranging from seizure detection wave analysis. The existing works leveraging self-supervised on EEG modeling mainly focus pretraining upon each individual dataset corresponding single...
The era of the Industrial Internet Things has led to an escalating menace Cyber-Physical Manufacturing Systems (CPMS) cyber-attacks. Presently, field intrusion detection for CPMS significant advancements. However, current methodologies require costs collecting historical data train models, which are tailored specific machining scenarios. Evolving scenarios in real world challenge adaptability these methods. In this paper, We found that code contains a complete process, is excellent basis....
Modeling continuous-time dynamics on irregular time series is critical to account for data evolution and correlations that occur continuously. Traditional methods including recurrent neural networks or Transformer models leverage inductive bias via powerful architectures capture complex patterns. However, due their discrete characteristic, they have limitations in generalizing paradigms. Though ordinary differential equations (Neural ODEs) variants shown promising results dealing with...
Unsupervised domain adaptation uses source data from different distributions to solve the problem of classifying unlabeled target domains. However, conventional methods require access data, which often raise concerns about privacy. In this paper, we consider a more practical but challenging setting where is unavailable and unlabeled. Specifically, address discrepancy perspective contrastive learning. The key idea our work learn domain-invariant feature by 1) performing clustering directly in...
Searchable encryption (SE) has emerged as a cryptographic primitive that allows data users to search on encrypted data. Most existing SE schemes usually delegate operations an intermediary such cloud server, which would inevitably result in single-point failure, privacy leakage, and even untrustworthy results. Several blockchain-based have been proposed alleviate these issues; however, they suffer from some issues, the support for multi-keyword multi-owner model, query storage availability....
With the prevalence of text-to-image generative models, their safety becomes a critical concern. adversarial testing techniques have been developed to probe whether such models can be prompted produce Not-Safe-For-Work (NSFW) content. However, existing solutions face several challenges, including low success rate and inefficiency. We introduce Groot, first automated framework leveraging tree-based semantic transformation for models. Groot employs decomposition sensitive element drowning...
Hashimoto's thyroiditis (HT) is an autoimmune disorder with unclear molecular mechanisms. While current diagnosis well-established, understanding of the gut-thyroid axis in HT remains limited. This study aimed to uncover novel signatures by integrating gut metagenome and host transcriptome data (miRNA/mRNA), potentially elucidating disease pathogenesis identifying new therapeutic targets. We recruited 31 early patients 30 healthy controls a two-stage (discovery validation). Blood fecal...