A5101713580- Topic Modeling
- Natural Language Processing Techniques
- Text and Document Classification Technologies
- Multimodal Machine Learning Applications
- Network Security and Intrusion Detection
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Web visibility and informetrics
- Photovoltaic System Optimization Techniques
- scientometrics and bibliometrics research
- Security in Wireless Sensor Networks
- Advanced Graph Neural Networks
- Energy Load and Power Forecasting
- Mobile Agent-Based Network Management
- Smart Grid Security and Resilience
- Geochemistry and Elemental Analysis
- Semantic Web and Ontologies
- Solar Radiation and Photovoltaics
- Geological and Geochemical Analysis
- Software-Defined Networks and 5G
- Advanced Image and Video Retrieval Techniques
- Speech and dialogue systems
- Sentiment Analysis and Opinion Mining
- Advanced Text Analysis Techniques
- Domain Adaptation and Few-Shot Learning
East China University of Science and Technology
2024
National University of Defense Technology
2023-2024
Dalian University of Technology
2021
The prediction method based on the partition of weather types has been widely used in photovoltaic prediction. Here, temporal characteristics PV data is studied, physical meaning sequential difference sequences analysed, and a characteristic index system including proposed. are partitioned self-organising map neural network algorithm. Case study shows that helpful to improve discrimination accuracy.
We propose a generalized method for boosting the generalization ability of pre-trained vision-language models (VLMs) while fine-tuning on downstream few-shot tasks. The idea is realized by exploiting out-of-distribution (OOD) detection to predict whether sample belongs base distribution or novel and then using score generated dedicated competition based scoring function fuse zero-shot classifier. fused classifier dynamic, which will bias towards if more likely from on, leading improved...
Cache-based approaches stand out as both effective and efficient for adapting vision-language models (VLMs). Nonetheless, the existing cache model overlooks three crucial aspects. 1) Pre-trained VLMs are mainly optimized image-text similarity, neglecting importance of image-image leading to a gap between pre-training adaptation. 2) The current is based on Nadaraya-Watson (N-W) estimator, which disregards intricate relationships among training samples while constructing weight function. 3)...
The advent of pre-trained vision-language foundation models has revolutionized the field zero/few-shot (i.e., low-shot) image recognition. key challenge to address under condition limited training data is how fine-tune in a parameter-efficient manner. Previously, numerous approaches tackling this have been proposed. Meantime, few survey papers are also published summarize these works. However, there still lacks unified computational framework integrate existing methods together, identify...
ABSTRACT This article deals with the design and defense issues of false data injection attack in distributed state estimation. At each time step, sensor sends its estimate to neighboring sensors. To verify security received data, a column vector composed differences is used as detection variable. Based on directly obtained by attacker wireless communication network, an strategy established degrade system estimation performance while not being detected detector. Under strategy, update error...
摘要: 铜厂铜-铁矿床在成矿时代、成矿物质来源及矿床成因等方面存在较大争议，限制了其成矿模式的建立以及进一步的找矿实践. 利用黄铜矿Re-Os同位素对该矿床进行定年，并利用LA-MC-ICP-MS技术对黄铜矿、黄铁矿及磁黄铁矿等硫化物开展原位硫同位素研究. 分析结果显示，5件黄铜矿Re-Os同位素等时线年龄为484±34 Ma（MSWD=8.7），表明铜厂铜-铁矿床形成于早古生代加里东期. 铜厂铜-铁矿床上部铜矿床中黄铜矿（+9.75‰~+13.1‰）和黄铁矿（+9.22‰~+13.9‰）的δ34S值略高于下部铁矿床中黄铜矿（+8.66‰~+10.9‰）、黄铁矿（+8.85‰~+11.0‰）和磁黄铁矿（+7.93‰~+9.28‰）. 计算得到早期成矿热液的δ34S∑S值约为+10.6‰，晚期成矿热液的δ34S∑S值约为+12.3‰，说明矿床硫是地幔硫混染海水硫形成的，热化学还原在海水硫酸盐还原过程中起到关键作用....
Event detection is an important subtask of information extraction, aiming to identify triggers and recognize event types in text. Previous state-of-the-art studies using graph neural networks (GNNs) are mainly applied obtain long distance features text have achieved impressive performance. However, these methods face the issues over-smoothing semantic feature destruction, when containing multiple GNN layers. For reasons, this paper proposes improved model for detection. The first a stacked...