A5025958247- Domain Adaptation and Few-Shot Learning
- Text and Document Classification Technologies
- Advanced Battery Technologies Research
- Advanced Graph Neural Networks
- Complex Network Analysis Techniques
- Face and Expression Recognition
- Data Stream Mining Techniques
- Advancements in Battery Materials
- Reliability and Maintenance Optimization
- Multimodal Machine Learning Applications
- Neural Networks and Applications
- Advanced Image and Video Retrieval Techniques
- Machine Learning and Data Classification
- Electric Vehicles and Infrastructure
- Artificial Intelligence in Healthcare
- Advanced Computing and Algorithms
- Opinion Dynamics and Social Influence
- Imbalanced Data Classification Techniques
- Anomaly Detection Techniques and Applications
- Computational Drug Discovery Methods
- Music and Audio Processing
- Advanced Bandit Algorithms Research
- Machine Learning and ELM
- Advanced Memory and Neural Computing
- Online Learning and Analytics
University of Electronic Science and Technology of China
2018-2024
Huzhou University
2022-2024
Quzhou University
2024
The development of a machine-learning method with high accuracy, generalization, and strong robustness for evaluating battery health states is essential in the field management. In this work, data-driven stacking regressor (SR) two-layer diagnostic framework was proposed to estimate state (SOH) predict remaining useful life (RUL). Five individual estimators were merged first layer, including bagging, gradient boosting regression (GBR), support vector (SVR), Hist-GBR, AdaBoost, linear (LR)...
Distributed data stream mining has gained increasing attention in recent years since many organizations collect tremendous amounts of streaming from different locations. Existing studies mainly focus on learning evolving concepts distributed streams, while the privacy issue is little investigated. In this article, for first time, we develop a federated framework concept-drifting called FedStream. The proposed method allows capturing by dynamically maintaining set prototypes with error-driven...
Synchronization is a ubiquitous phenomenon in nature that enables the orderly presentation of information. In human brain, for instance, functional modules such as visual, motor, and language cortices form through neuronal synchronization. Inspired by biological brains previous neuroscience studies, we propose an interpretable neural network incorporating synchronization mechanism. The basic idea to constrain each neuron, convolution filter, capture single semantic pattern while...
The dynamic nature of open-world scenarios has attracted more attention to class incremental learning (CIL). However, existing CIL methods typically presume the availability complete ground-truth labels throughout training process, an assumption rarely met in practical applications. Consequently, this paper explores a challenging problem unsupervised (UCIL). essence addressing lies effectively capturing comprehensive feature representations and discovering unknown novel classes. To achieve...
As the growing market shared by electric vehicles, safety accidents of vehicles caused lithium-ion battery failures has also increased. Hence, early detection faults becomes critical to ensure life and property safety. In this study, a fault method leveraging feature fusion LOF algorithm was proposed. By using normalized voltage optimal sliding window length, can effectively identify subtle changes faulty. The reliability effectiveness proposed were confirmed through analysis real-world...
Few-shot class-incremental learning (FSCIL) is a step forward in the realm of incremental learning, catering to more realistic context. In typical scenarios, initial session possesses ample data for effective training. However, subsequent sessions often lack sufficient data, leading model face simultaneously with challenges catastrophic forgetting and overfitting few-shot learning. Existing methods employ complex maintain balance plasticity stability. this study, we break design lazy...
Multi-label classification problem has gained growing attention in recent years due to its diverse applications real-world problems such as image annotation and query suggestions. However, traditional multi-label methods tend fail the inconsistency between input output space, where similar instances feature space may have distinct semantic labels space. To eliminate problem, this paper, we propose a supervised metric learning approach for classification, called MLMLI, which attempts learn...