A5100683309- Topic Modeling
- Natural Language Processing Techniques
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Speech and Audio Processing
- Land Use and Ecosystem Services
- Domain Adaptation and Few-Shot Learning
- Cryptographic Implementations and Security
- Advanced Text Analysis Techniques
- Physical Unclonable Functions (PUFs) and Hardware Security
- Expert finding and Q&A systems
- Emotion and Mood Recognition
- Sentiment Analysis and Opinion Mining
- Advanced Image and Video Retrieval Techniques
- Face recognition and analysis
- Water Resources and Sustainability
- Advanced Image Processing Techniques
- COVID-19 diagnosis using AI
- Platelet Disorders and Treatments
- Sustainable Building Design and Assessment
- Intimate Partner and Family Violence
- Service-Oriented Architecture and Web Services
- Sport and Mega-Event Impacts
- Advanced Malware Detection Techniques
- Cultural Industries and Urban Development
Tokyo Institute of Technology
2024-2025
Nanyang Technological University
2022-2024
Nanjing University of Finance and Economics
2024
Taizhou Vocational and Technical College
2023
Huzhou Vocational and Technical College
2023
Nanjing Normal University
2023
Aerospace Center Hospital
2023
Zhejiang A & F University
2023
Gansu Institute of Political Science and Law
2020
Recently, personality trait recognition, which aims to identify people's first impression behavior data and analyze psychological characteristics, has been an interesting active topic in psychology, affective neuroscience artificial intelligence. To effectively take advantage of spatio-temporal cues audio-visual modalities, this paper proposes a new method multimodal recognition integrating modalities based on hybrid deep learning framework, is comprised convolutional neural networks (CNN),...
As an exquisite asset of Chinese traditional culture, opera occupies a place high esteem within the world’s cultural and artistic treasury. The impact emerging cultures has threatened future necessitating thorough examination historical context Grand Canal opera. There is insufficient research on spatial evolution culture along Canal; thus, this study takes ancient stages, representative relic opera, as entry point employs methods such kernel density analysis standard deviation ellipse to...
With the implementation of China’s rural revitalization strategy, construction low-pollution villages has become an urgent demand for residents to achieve ecologically livable environment. This paper aims construct a new evaluation system “zero-pollution village” inspired by principle CASBEE. A village is regarded as relatively independent ecosystem, and designed improve “internal environmental quality Q” while reducing “external load L”. Thus, included two target strata (Q L) 3 layers...
Stroke is considered one of the most common and life-threatening manifestations antiphospholipid syndrome (APS), which leads to high mortality permanent disability. This study investigated prevalence potential risk factors stroke in APS. We enrolled 361 APS patients retrospectively from 2009 2022 at Peking University People’s Hospital. was found 25.8% (93/361) participants. The multivariate logistic regression showed that hypertension, diabetes, livedo reticularis, other central nervous...
Mask-face detection has been a significant task since the outbreak of COVID-19 pandemic in early 2020. While various reviews on mask-face techniques up to 2021 are available, little reviewed distinction between two-class (i.e., wearing mask and without mask) three-class masking, which includes an additional incorrect-mask-wearing class. Moreover, no formal review conducted implementing models hardware systems or mobile devices. The objectives this paper three-fold. First, we aimed provide...
Hybrid Question-Answering (HQA), which targets reasoning over tables and passages linked from table cells, has witnessed significant research in recent years. A common challenge HQA other passage-table QA datasets is that it generally unrealistic to iterate all rows, columns, retrieve evidence. Such a made difficult for previous studies show their ability retrieving answers. To bridge this gap, we propose novel Table-alignment-based Cell-selection Reasoning model (TACR) hybrid text QA,...
Continual learning (CL) with Vision-Language Models (VLMs) has overcome the constraints of traditional CL, which only focuses on previously encountered classes. During CL VLMs, we need not to prevent catastrophic forgetting incrementally learned knowledge but also preserve zero-shot ability VLMs. However, existing methods require additional reference datasets maintain such and rely domain-identity hints classify images across different domains. In this study, propose Regression-based...
Hybrid Question-Answering (HQA), which targets reasoning over tables and passages linked from table cells, has witnessed significant research in recent years. A common challenge HQA other passage-table QA datasets is that it generally unrealistic to iterate all rows, columns, retrieve evidence. Such a made difficult for previous studies show their ability retrieving answers. To bridge this gap, we propose novel Table-alignment-based Cell-selection Reasoning model (TACR) hybrid text QA,...
Correlation Optimization with Deep Learning Analysis (CO-DLA) is a state-of-the-art non-profiling side-channel analysis based on Learning. In this paper, we proposes novel Wavelet Scattering Transform-based CO-DLA (WST-CO-DLA). WST-CO-DLA utilizes the non-training-required Transform (WST) to extract time-frequency features for CO-DLA, thus offering three main contributions. Firstly, WST eliminates need training-required convolution blocks in CNN-based or reduces number of linear layers...
Currently, the security issues of Web services are hot area in information system (IS). This research mainly discusses key technologies access control focusing on following works: After analyzing dynamic characteristic application nature for services, a Dynamic Role-Based Access Control using Context and Trust model (abbreviated as CT-DRBAC) is proposed. During both subject invoking request object providing service resources nature. So, policies needed to consider The proposed has been...
To break the three lockings during backpropagation (BP) process for neural network training, multiple decoupled learning methods have been investigated recently. These either lead to significant drop in accuracy performance or suffer from dramatic increase memory usage. In this paper, a new form of learning, named training scheme with re-computation and weight prediction (DTRP) is proposed. DTRP, adopted solve explosion problem, proposed deal delay caused by re-computation. Additionally,...
Multimodal Emotion Recognition (MER) has recently become a popular and challenging topic. The most key challenge in MER is how to effectively fuse multimodal information. Most existing works may fail consider the cross-modal interaction information for MER. To address this problem, paper proposes new framework based on audio text by using Hybrid Attention Networks (MER-HAN). proposed MER-HAN combines three different attention mechanisms such as local intra-modal attention, global inter-modal...
Hybrid Question-Answering (HQA), which targets reasoning over tables and passages linked from table cells, has witnessed significant research in recent years. A common challenge HQA other passage-table QA datasets is that it generally unrealistic to iterate all rows, columns, retrieve evidence. Such a made difficult for previous studies show their ability retrieving answers. To bridge this gap, we propose novel Table-alignment-based Cell-selection Reasoning model (TACR) hybrid text QA,...