A5056134483- Traffic Prediction and Management Techniques
- Air Quality and Health Impacts
- Air Quality Monitoring and Forecasting
- Geographic Information Systems Studies
- Advanced Image and Video Retrieval Techniques
- Handwritten Text Recognition Techniques
- Web Data Mining and Analysis
- Transportation Planning and Optimization
- Image Processing and 3D Reconstruction
- Network Security and Intrusion Detection
- Image Retrieval and Classification Techniques
- Human Mobility and Location-Based Analysis
- Anomaly Detection Techniques and Applications
- Noise Effects and Management
- 3D Modeling in Geospatial Applications
- Vehicle emissions and performance
- Complex Network Analysis Techniques
- Context-Aware Activity Recognition Systems
- Reinforcement Learning in Robotics
- Simulation Techniques and Applications
- Advanced Text Analysis Techniques
- Infrastructure Maintenance and Monitoring
- Model Reduction and Neural Networks
- Meteorological Phenomena and Simulations
- Time Series Analysis and Forecasting
Twin Cities Orthopedics
2022-2024
University of Minnesota
2022-2024
University of Minnesota System
2021-2023
Sichuan International Studies University
2023
Institute of Software
2022
Chinese Academy of Sciences
2022
University of Chinese Academy of Sciences
2022
University of Southern California
2017-2021
Chongqing University of Posts and Telecommunications
2020
Southern California University for Professional Studies
2018
Forecasting spatially correlated time series data is challenging because of the linear and non-linear dependencies in temporal spatial dimensions. Air quality forecasting one canonical example such tasks. Existing work, e.g., auto-regressive integrated moving average (ARIMA) artificial neural network (ANN), either fails to model dependency or cannot effectively consider relationships between multiple data. In this paper, we present an approach for short-term PM2.5 concentrations using a deep...
Major societal and environmental challenges involve complex systems that have diverse multi-scale interacting processes. Consider, for example, how droughts water reserves affect crop production agriculture industrial needs quality availability. Preventive measures, such as delaying planting dates adopting new agricultural practices in response to changing weather patterns, can reduce the damage caused by natural Understanding these human processes one another allows forecasting effects of...
Air quality models are important for studying the impact of air pollutant on health conditions at a fine spatiotemporal scale. Existing work typically relies area-specific, expert-selected attributes pollution emissions (e,g., transportation) and dispersion (e.g., meteorology) building model each combination study areas, types, scales. In this paper, we present data mining approach that utilizes publicly available OpenStreetMap (OSM) to automatically generate an concentrations particulate...
Time-resolved quantification of physical activity can contribute to both personalized medicine and epidemiological research studies, for example, managing identifying triggers asthma exacerbations. A growing number reportedly accurate machine learning algorithms human recognition (HAR) have been developed using data from wearable devices (eg, smartwatch smartphone). However, many HAR depend on fixed-size sampling windows that may poorly adapt real-world conditions in which bouts are unequal...
Text on historical maps contains valuable information providing georeferenced historical, political, and cultural contexts. However, text extraction from has been challenging due to the lack of (1) effective methods (2) training data. Previous approaches use ad-hoc steps tailored only specific map styles. Recent machine learning-based spotters (e.g., for scene images) have potential solve these challenges because their flexibility in supporting various types instances. remain extracting...
Scanned historical maps in libraries and archives are valuable repositories of geographic data that often do not exist elsewhere. Despite the potential machine learning tools like Google Vision APIs for automatically transcribing text from these into machine-readable formats, they work well with large-sized images (e.g., high-resolution scanned documents), cannot infer relation between recognized other datasets, challenging to integrate post-processing tools. This paper introduces mapKurator...
Many scientific prediction problems have spatiotemporal data- and modeling-related challenges in handling complex variations space time using only sparse unevenly distributed observations. This paper presents a novel deep learning architecture, Deep predictions for LocATion-dependent Time-sEries data (DeepLATTE), that explicitly incorporates theories of spatial statistics into neural networks to addresses these challenges. In addition feature selection module module, DeepLATTE contains an...
With the widespread installation of location-enabled devices on public transportation, vehicles are generating massive amounts trajectory data in real time. However, using these for meaningful analysis requires careful considerations storing, managing, processing, and visualizing data. Using location Los Angeles Metro bus system, along with publicly available schedule data, we conduct a processing analyses study to measure performance transportation system utilizing number metrics including...
Recent advances in incorporating physical knowledge into deep neural networks can estimate previously unknown governing partial differential equations (PDEs) a data-driven way. They have shown promising results spatiotemporal predictive learning. However, these methods typically assume universal PDEs across space, which is impractical for modeling complex phenomena with high spatial variability (e.g., climate). Also, they cannot effectively model the evolution of potential errors estimating...
Abstract Compared with natural images, geospatial images cover larger area and have more complex image contents. There are few algorithms for generating controllable their results of low quality. In response to this problem, paper proposes Geospatial Style Generative Adversarial Network generate high‐quality images. Current conditional generators suffer the mode collapse problem in field. The is addressed via a modified seeking regularization term contrastive learning theory. Besides,...
Large network logs, recording multivariate time series generated from heterogeneous devices and sensors in a network, can reveal important information about abnormal activities, such as intrusions packet losses. Existing machine learning methods for anomaly detection on multiple typically assume that 1) infrequent behaviors beyond some inference threshold are anomalous unsupervised models or 2) require large set of labeled normal sequences supervised models. However, practice, the reported...
There are increasing numbers of online sources real-time and historical location-dependent time-series data describing various types environmental phenomena, e.g., traffic conditions air quality levels. When coupled with the information that characterizes natural built environments, these can help better understand interactions between within human social systems ecosystem. Nevertheless, still limited by their spatial temporal resolution for downstream use (e.g., generating residential-level...
Large network logs, recording multivariate time series generated from heterogeneous devices and sensors in a network, can often reveal important information about abnormal activities, such as intrusions device malfunctions. Existing machine learning methods for anomaly detection on typically assume that 1) normal sequences would have consistent behavior training unsupervised models, or 2) require large set of labeled supervised models. However, practice, activities demonstrate significantly...
Scanned historical maps in libraries and archives are valuable repositories of geographic data that often do not exist elsewhere. Despite the potential machine learning tools like Google Vision APIs for automatically transcribing text from these into machine-readable formats, they work well with large-sized images (e.g., high-resolution scanned documents), cannot infer relation between recognized other datasets, challenging to integrate post-processing tools. This paper introduces mapKurator...
This paper focuses on the issues of rural education development in China. Education is cornerstone development. As a populous and developing country, China pays great attention to issues. The Chinese government people have made many directional efforts for education, but there are still regional, cultural, linguistic problems practical stages policy implementation due China's vast territory. Especially aspect problems, revival also promotes revitalization. analyzes introduces current...
<sec> <title>BACKGROUND</title> Time-resolved quantification of physical activity can contribute to both personalized medicine and epidemiological research studies, for example, managing identifying triggers asthma exacerbations. A growing number reportedly accurate machine learning algorithms human recognition (HAR) have been developed using data from wearable devices (eg, smartwatch smartphone). However, many HAR depend on fixed-size sampling windows that may poorly adapt real-world...