A5008871638- Anomaly Detection Techniques and Applications
- Time Series Analysis and Forecasting
- Smart Grid and Power Systems
- Brain Tumor Detection and Classification
- Complex Network Analysis Techniques
- Sports Analytics and Performance
- Pesticide Residue Analysis and Safety
- Biomedical Text Mining and Ontologies
- Advanced Sensor and Control Systems
- Fault Detection and Control Systems
- Topic Modeling
- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Wind and Air Flow Studies
- Machine Learning and Algorithms
- Artificial Intelligence in Healthcare
- Artificial Immune Systems Applications
- Spectroscopy and Chemometric Analyses
- Infection Control and Ventilation
- Imbalanced Data Classification Techniques
- Computational Physics and Python Applications
- Advanced Graph Neural Networks
- Data-Driven Disease Surveillance
- Autonomous Vehicle Technology and Safety
- Advanced Algorithms and Applications
University of Notre Dame
2022-2025
Guilin University of Electronic Technology
2023
With the increasing deployment of small unmanned aerial systems (sUASs) on various tasks, it becomes crucial to analyze and detect anomalies from their flight logs. To support research in this area, we curate Drone Log Anomaly (DLA), first real-world time series anomaly detection dataset domain sUASs, which contains 41 sUAS logs annotated with types anomalies. As tend occur low-density areas within a distribution, propose graphical normalizing flows (GNF), graph-based autoregressive deep...
CyberPhysical systems (CPS) must be closely monitored to identify and potentially mitigate emergent problems that arise during their routine operations. However, the multivariate time-series data which they typically produce can complex understand analyze. While formal product documentation often provides example plots with diagnostic suggestions, sheer diversity of attributes, critical thresholds, interactions overwhelming non-experts who subsequently seek help from discussion forums...
The spread of COVID-19 throughout the world has led to cataclysmic consequences on global community, which poses an urgent need accurately understand and predict trajectories pandemic. Existing research relied graph-structured human mobility data for task pandemic forecasting. To perform forecasting in United States, we curate Large-MG, a large-scale dataset that contains 66 dynamic graphs, with each graph having over 3k nodes average 540k edges. One drawback existing Graph Neural Networks...
Learning effective recipe representations is essential in food studies. Unlike what has been developed for image-based retrieval or learning structural text embeddings, the combined effect of multi-modal information (i.e., images, text, and relation data) receives less attention. In this paper, we formalize problem representation to integrate visual, textual, relational into embeddings. particular, first present Large-RG, a new graph data with over half million nodes, making it largest date....
Artificial neural networks are widely used in various fields, such as intelligent road networks, Internet of Things, and smart medical systems due to their ability process large amounts data parallel, store information a distributed manner, self-organize self-learn. Cloud computing technology has further expanded the development network applications. However, user often contains sensitive information, once management right is transferred cloud, it faces serious security privacy issues. In...
View Video Presentation: https://doi.org/10.2514/6.2022-3540.vid The rapid growth in the number of small Unmanned Aerial Systems (sUAS) and related increase reported airspace violations incidents, makes it imperative to improve sUAS safety. In this paper, we describe our vision preliminary results for deploying data analytic solutions onboard order detect, diagnose, potentially mitigate emergent anomalies. We take a multi-prong approach support both run-time post-mortem analysis flight use...
The yield of a chemical reaction quantifies the percentage target product formed in relation to reactants consumed during reaction. Accurate prediction can guide chemists toward selecting high-yield reactions synthesis planning, offering valuable insights before dedicating time and resources wet lab experiments. While recent advancements have led overall performance improvement across entire range, an open challenge remains enhancing predictions for reactions, which are greater concern...
Recently, Wordle is a daily Scrabble game provided by the New York Times, which has been loved many players. Based on this game, paper focuses predicting based neural network models, using methods such as One-Hot Encoding, correlation test, and so to solve specific task divided into four major problems. For problem one, firstly, given data are preprocessed, some outliers eliminated, while original replaced mean value of number people, filtered normalized range [0, 1]. Then, Nested Two-Layer...