A5100328763- Context-Aware Activity Recognition Systems
- Data Visualization and Analytics
- Privacy-Preserving Technologies in Data
- Indoor and Outdoor Localization Technologies
- Video Coding and Compression Technologies
- Data Management and Algorithms
- Image Enhancement Techniques
- Image and Video Quality Assessment
- Data Stream Mining Techniques
- Fish Ecology and Management Studies
- Advanced Image and Video Retrieval Techniques
- Scientific Computing and Data Management
- Modular Robots and Swarm Intelligence
- Advanced Neural Network Applications
- Distributed systems and fault tolerance
- Molecular Communication and Nanonetworks
- Ethics and Social Impacts of AI
- Hydrology and Sediment Transport Processes
- Biomimetic flight and propulsion mechanisms
- Multimodal Machine Learning Applications
- Adversarial Robustness in Machine Learning
- Hydrological Forecasting Using AI
- Neural dynamics and brain function
- Opportunistic and Delay-Tolerant Networks
- IoT and Edge/Fog Computing
Kunming Medical University
2024
China Institute of Water Resources and Hydropower Research
2023-2024
Northwestern Polytechnical University
2022-2024
The Ohio State University
2021
Zhejiang University
2020
University of San Francisco
2020
Chinese Academy of Agricultural Engineering
2019
National University of Singapore
2006
The rise of mobile devices with abundant sensory data and local computing capabilities has driven the trend federated learning (FL) on these devices. And personalized FL (PFL) emerges to train specific deep models for each device address heterogeneity varying performance preferences. However, training times vary significantly, resulting in either delay (when waiting slower aggregation) or accuracy decline aggregation proceeds without waiting). In response, we propose a shift towards...
In the fish passage facility design, understanding coupled effects of hydrodynamics on behaviour is particularly important. The flow field caused by movement however are usually obtained via time-consuming transient numerical simulation. Hence, a hybrid deep neural network (HDNN) approach designed to predict unsteady around fish. basic architecture HDNN includes UNet convolution (UConv) module and bidirectional convolutional long-short term memory (BiConvLSTM) module. Specifically, UConv...
PDAs, cellular phones and other mobile devices are now capable of supporting complex data manipulation operations. Here, we focus on ad-hoc spatial joins datasets residing in multiple non-cooperative servers. Assuming that there is no mediator available, the must be evaluated device. Contrary to common applications consider cost at server side, our main issue minimization transferred data, while meeting resource constraints We show existing methods, based partitioning pruning, inadequate...
Abstract Significant increases in suspended sediment concentration (SSC) always occur during dam removal or flushing processes, which could result acute impacts on aquatic organisms. An analysis of the potential impact Huangheyuan hydropower upper reaches Yellow River is required for adaptive management. But reports high SSC fishes living rivers plateau areas are limited. Further studies factors, fish behavior, and assessment methods were performed this paper. The response a typical fish,...
The ubiquity of camera-embedded devices and the advances in deep learning have stimulated various intelligent mobile video applications. These applications often demand on-device processing streams to deliver real-time, high-quality services for privacy robustness concerns. However, performance these is constrained by raw streams, which tend be taken with small-aperture cameras ubiquitous platforms dim light. Despite extensive low-light enhancement solutions, they are unfit deployment due...
The rise of mobile devices equipped with numerous sensors, such as LiDAR and cameras, has spurred the adoption multi-modal deep intelligence for distributed sensing tasks, smart cabins driving assistance. However, arrival times sensory data vary due to modality size network dynamics, which can lead delays (if waiting slower data) or accuracy decline inference proceeds without waiting). Moreover, diversity dynamic nature systems exacerbate this challenge. In response, we present a shift...
PDAs, cellular phones and other mobile devices are now capable of supporting complex data manipulation operations. Here, we focus on ad-hoc spatial joins datasets residing in multiple non-cooperative servers. Assuming that there is no mediator available, the must be evaluated device. Contrary to common applications consider cost at server side, our main issue minimization transferred data, while meeting resource constraints We show existing methods, based partitioning pruning, inadequate...
Knowledge Graph Embedding (KGE) is a fundamental technique that extracts expressive representation from knowledge graph (KG) to facilitate diverse downstream tasks. The emerging federated KGE (FKGE) collaboratively trains distributed KGs held among clients while avoiding exchanging clients' sensitive raw KGs, which can still suffer privacy threats as evidenced in other model trainings (e.g., neural networks). However, quantifying and defending against such remain unexplored for FKGE...
Camera-equipped devices and deep learning advancements have driven the development of intelligent mobile video apps. These apps require on-device processing streams for real-time, high-quality services while addressing privacy robustness. However, their performance is limited by low-light conditions small-aperture cameras in platforms. Existing enhancement solutions are unsuitable due to complex models lack energy efficiency. We introduce MoEnlight, an energy-conscious system enhancing on...
As worldwide capability to collect, store, and manage information continues grow, the resulting datasets become increasingly difficult understand extract insights from. Interactive data visualizations offers a promising avenue efficiently navigate gain from highly complex datasets, but velocity of modern streams often means that precomputed representations or summarizations will quickly obsolete. Our system, Agami, provides live-updating, interactive over streaming data. We leverage...
Deep neural networks are proved to be very effective solve problems on image classification, object detection and segmentation. However, in cases where only limited hardware is acquired, it may a problem deploy big models with excellent performance as they sometimes calculation consuming. To overcome the limits power, memory calculation, channel pruning proposed compress model wise soon become common approach have compressed. Generally, three-stage pipeline containing training, finetuning....
Training long-horizon robotic policies in complex physical environments is essential for many applications, such as manipulation. However, learning a policy that can generalize to unseen tasks challenging. In this work, we propose achieve one-shot task generalization by decoupling plan generation and execution. Specifically, our method solves three steps: build paired abstract environment simplifying geometry physics, generate trajectories, solve the original an abstract-to-executable...