A5101445040- Advanced Image and Video Retrieval Techniques
- Topic Modeling
- Speech and dialogue systems
- Natural Language Processing Techniques
- Image Retrieval and Classification Techniques
- Video Surveillance and Tracking Methods
- Handwritten Text Recognition Techniques
- Anomaly Detection Techniques and Applications
- Digital Marketing and Social Media
- Video Analysis and Summarization
- Blind Source Separation Techniques
- High-Energy Particle Collisions Research
- Sparse and Compressive Sensing Techniques
- Neural Networks and Applications
- Particle physics theoretical and experimental studies
- Quantum Chromodynamics and Particle Interactions
- Context-Aware Activity Recognition Systems
- Metaheuristic Optimization Algorithms Research
- Human Pose and Action Recognition
- Complex Network Analysis Techniques
- Hand Gesture Recognition Systems
- Vehicle License Plate Recognition
- Text and Document Classification Technologies
- Time Series Analysis and Forecasting
- Intelligent Tutoring Systems and Adaptive Learning
University of Wollongong
2016-2025
Chongqing University
2012-2025
Chongqing Cancer Hospital
2025
Wuhan University of Technology
2006-2024
Southeast University
2023
Henan Cancer Hospital
2023
Zhengzhou University
2023
Wenzhou University
2023
Sichuan University
2023
Shanghai Jiao Tong University
2006-2022
The authors present a real-time face tracker. system has achieved rate of 30+ frames/second using an HP-9000 workstation with frame grabber and Canon VC-Cl camera. It can track person's while the person moves freely (e.g., walks, jumps, sits down stands up) in room. Three types models have been employed developing system. First, they stochastic model to characterize skin color distributions human faces. information provided by is sufficient for tracking various poses views. This adaptable...
A person seeking someone else's attention is normally able to quickly assess how interruptible they are. This assessment allows for behavior we perceive as natural, socially appropriate, or simply polite. On the other hand, today's computer systems are almost entirely oblivious human world operate in, and typically have no way take into account interruptibility of user. paper presents a Wizard Oz study exploring whether, how, robust sensor-based predictions might be constructed, which...
To successfully interact with and learn from humans in cooperative modes, robots need a mechanism for recognizing, characterizing, emulating human skills. In particular, it is our interest to develop the recognizing simple actions, i.e., activity manual operation where no sensory feedback available. this end, we have developed method model such actions using hidden Markov (HMM) representation. We proposed an approach address two critical problems action modeling: classifying action-intent,...
Graph-based semi-supervised learning has gained considerable interests in the past several years thanks to its effectiveness combining labeled and unlabeled data through label propagation for better object modeling classification. A critical issue constructing a graph is weight assignment where of an edge specifies similarity between two points. In this paper, we present novel technique measure similarities among points by decomposing each point as L <sub...
Abstract As an essential component of human cognition, cause–effect relations appear frequently in text, and curating from text helps building causal networks for predictive tasks. Existing causality extraction techniques include knowledge-based, statistical machine learning (ML)-based, deep learning-based approaches. Each method has its advantages weaknesses. For example, knowledge-based methods are understandable but require extensive manual domain knowledge have poor cross-domain...
Cyber attacks and intrusions have become the major obstacles to adoption of Industrial Internet Things (IIoT) in critical industries. Imbalanced data distribution is a common problem IIoT environments that negatively influence machine learning-based intrusion detection systems (IDSs). To address this issue, we introduce EvolCostDeep, hybrid model stacked autoencoders (SAE) convolutional neural networks (CNNs) with new cost-dependent loss function. The function aims optimize model's...
Context-sensing for context-aware HCI challenges the traditional sensor fusion methods with dynamic configuration and measurement requirements commensurate human perception. The Dempster-Shafer theory of evidence has uncertainty management inference mechanisms analogous to our reasoning process. Our Sensor Fusion Context-aware Computing Project aims build a generalizable architecture in systematic way. This naturally leads us choose approach as first implementation algorithm paper discusses...
In this paper, we present our efforts towards developing an intelligent tourist system. The system is equipped with a unique combination of sensors and software. hardware includes two computers, GPS receiver, lapel microphone plus earphone, video camera head-mounted display. This multimodal interface to take advantage speech gesture input provide assistance for tourist. software supports natural language processing, recognition, machine translation, handwriting recognition fusion. A vision...
We present a new adaptive algorithm for automatic detection of text from natural scene. The initial cues regions are first detected the captured image/video. An color modeling and searching is then utilized near cues, to discriminate text/non-text regions. EM optimization used modeling, under constraint layout relations specific language. proposed combines advantages several previous approaches detection, utilizes focus-of-attention approach finding. whole applied in prototype system that...
Existing salient object detection models favor over-segmented regions upon which saliency is computed. Such local are less effective on representing holistically and degrade emphasis of entire objects. As a result, the existing methods often fail to highlight an in complex background. Toward better grouping objects background, this paper, we consider graph cut, more specifically, normalized cut (Ncut) for detection. Since Ncut partitions energy minimization fashion, resulting eigenvectors...
Analysis of online user-generated content is receiving attention for its wide applications from both academic researchers and industry stakeholders. In this pilot study, we address common Big Data problems time constraints memory costs involved with using standard single-machine hardware software. A novel processing framework proposed to investigate a niche subset popular culture on Douban, well-known Chinese-language social network. Huge data samples are harvested via an asynchronous...
Solid organ transplantation (SOT) is vital for end-stage failure but faces challenges like shortage and rejection. Artificial intelligence (AI) offers potential to improve outcomes through better matching, success prediction, automation. However, the evolution of AI in SOT research remains underexplored. This study uses bibliometric analysis identify trends, hotspots, key contributors field. 821 articles from Web Science Core Collection were exported analysis. Microsoft Excel 2021 was used...
Maximum Mean Discrepancy (MMD) is widely used in a number of domain adaptation (DA) methods and shows its effectiveness aligning data distributions across domains. However, previous DA research, MMD-based focus mostly on distribution alignment, ignore to optimize the decision boundary for classification-aware DA, thereby falling short reducing upper error bound. In this paper, we propose strengthened MMD measurement, namely, Decision Boundary optimization-informed (DB-MMD), which enables...