A5085955054- Crystallization and Solubility Studies
- X-ray Diffraction in Crystallography
- Fault Detection and Control Systems
- Mineral Processing and Grinding
- Non-Destructive Testing Techniques
- Machine Fault Diagnosis Techniques
- Advanced Computational Techniques and Applications
- Imbalanced Data Classification Techniques
- Industrial Vision Systems and Defect Detection
- Topic Modeling
- Advanced MRI Techniques and Applications
- Multimodal Machine Learning Applications
- Advanced Decision-Making Techniques
- Crystallography and molecular interactions
- Cerebrovascular and Carotid Artery Diseases
- Oil and Gas Production Techniques
- Advanced Statistical Process Monitoring
- Domain Adaptation and Few-Shot Learning
- Machine Learning in Materials Science
- Machine Learning and Data Classification
- Spectroscopy and Chemometric Analyses
- Icing and De-icing Technologies
- Machine Learning and ELM
- Natural Language Processing Techniques
- Artificial Intelligence in Healthcare
Zhejiang University
2023-2025
University of Illinois Urbana-Champaign
2023-2025
Zhejiang University-University of Edinburgh Institute
2023-2025
Jiangsu University
2024
Chinese People's Armed Police Force
2024
Pingjin Hospital
2024
Xuzhou Cancer Hospital
2024
Southeast University
2023
Huazhong University of Science and Technology
2023
Tsinghua University
2019-2022
Fault diagnosis in an open world refers to the tasks that need cope with previously unknown faults online stage. It faces a great challenge yet be addressed—that is, data of may classified as normal samples high probability. In this article, we develop effective solution for by using supervised contrastive learning learn discriminative and compact embedding known situation fault situations. Specifically, addition contrasting given sample other instances is case conventional methods, our...
Class-incremental fault diagnosis requires a model to adapt new classes while retaining previous knowledge. However, limited research exists for imbalanced and long-tailed data. Extracting discriminative features from few-shot data is challenging, adding often demands costly retraining. Moreover, incremental training of existing methods risks catastrophic forgetting, severe class imbalance can bias the model's decisions toward normal classes. To tackle these issues, we introduce Supervised...
In this article, a new fault diagnosis problem is formulated, which involves large number of normal samples and in almost all the classes are few-shot classes. Although common many industrial scenarios, it remains challenge overlooked previous studies. To develop novel solution for addressing challenge, we employ long-tailed distribution work name task accordingly. Specifically, divide procedure into representation learning classification. On basis, propose method using progressively...
To explore the effects of propofol and ciprofol on patient euphoric reactions during sedation in patients undergoing gastroscopy to investigate potential factors that may influence patients.
Although convolution neural network (CNN) has achieved certain success in fault detection and diagnosis (FDD) tasks the chemical engineering industry, performance credibility of CNN-based FDD methods are greatly limited by two factors. First, CNN relies upon strong temporal/spatial correlation data, which is very difficult to obtain generic tabular data. Second, most have poor interpretability due encapsulation mechanism feature extraction, thus there great difficulty identifying root-cause...
In real industrial processes, fault diagnosis methods are required to learn from limited samples since the procedures mainly under normal conditions and faults rarely occur. Although attention mechanisms have become increasingly popular for task of diagnosis, existing attention-based still unsatisfying above practical applications. First, pure architectures like transformers need a substantial quantity offset lack inductive biases thus performing poorly samples. Moreover, poor classification...
Current deep-learning-based fault diagnosis methods, though proven to be successful with sufficient data, cannot well address the challenges of sample availability in real-world industrial scenarios. To this challenge, article proposes an effective approach by exploiting data generation and selection techniques. Specifically, we first develop a balancing generative adversarial network (BAGAN)-based technique generate more discriminative samples utilizing not only samples, but also normal...
As a method widely used in fault detection, principal component analysis (PCA) still has challenges applicability due to its sensitivity outliers and difficulty components (PCs) interpretation. In this paper, robust sparse PCA (RSPCA) model is proposed improve the robustness interpretability of PCA-based detection methods. Specifically, better achieved through capturing maximal L1-norm variance while non-zero loadings are given for PCs achieve improved interpretability. A developed based on...
Compared with traditional multivariate statistical techniques, deep neural networks have been frequently used for single-mode fault diagnosis and shown promising results. However, in the real world, a complex industrial process may several modes fewer samples than normal. Although techniques focus on multimode diagnosis, those methods usually identify then locally diagnose faults while ignoring relative information across data imbalance problems. In this paper, learning-based method...
Medical zero-shot relation triplet extraction, referred to as Med-ZeroRTE, requires the model extract triplets comprising entities and relations from medical sentences. Importantly, sentences include that were unseen during model's training phase. While Med-ZeroRTE had not been formally explored before this work, limited availability of datasets, influenced by privacy concerns annotation costs, emphasizes necessity exploring Med-ZeroRTE. This exploration faces two main challenges: Firstly,...
Fault diagnosis is crucial in monitoring machines within industrial processes. With the increasing complexity of working conditions and demand for safety during production, diverse methods are required, an integrated fault system capable handling multiple tasks highly desired. However, subtasks often studied separately, current still need improvement such a generalized system. To address this issue, we propose out-of-distribution (GOOFD) framework to integrate subtasks. Additionally, unified...
Abstract The presence of multiple failure severities in the wind turbine pitch system due to long-term wear and tear poses challenges accurately classifying system’s health condition, thus increasing maintenance costs or damage risks. This paper proposes a novel method based on hard sample mining (HSM)-enabled supervised contrastive learning address this problem. proposed leverages powerful feature extraction capabilities extract discriminative features from highly imbalanced data....
SummaryBackgroundThis study explores the potential of deep learning-based convolutional neural network (CNN) to automatically recognize MMD using MRA images from atherosclerotic disease (ASD) and normal control (NC).MethodsIn this retrospective in China, 600 participants (200 MMD, 200 ASD NC) were collected one institution as an internal dataset for training 60 another external testing set validation. All divided into (N = 450) validation sets 90), 60), 60). The input CNN models comprised...
In real industrial processes, fault diagnosis methods are required to learn from limited samples since the procedures mainly under normal conditions and faults rarely occur. Although attention mechanisms have become popular in field of diagnosis, existing attention-based still unsatisfying for above practical applications. First, pure architectures like transformers need a large number offset lack inductive biases thus performing poorly samples. Moreover, poor classification dilemma further...
Fault diagnosis is essential in industrial processes for monitoring the conditions of important machines. With ever-increasing complexity working and demand safety during production operation, different methods are required, more importantly, an integrated fault system that can cope with multiple tasks highly desired. However, subtasks often studied separately, currently available still need improvement such a generalized system. To address this issue, we propose Generalized...
Distant supervision reduces the reliance on human annotation in named entity recognition tasks. The class-level imbalanced distant is a realistic and unexplored problem, popular method of self-training can not handle learning. More importantly, dominated by high-performance class selecting candidates, deteriorates low-performance with bias generated pseudo label. To address imbalance performance, we propose class-rebalancing framework for improving distantly-supervised recognition. In...