A5100433703- Privacy-Preserving Technologies in Data
- Mobile Crowdsensing and Crowdsourcing
- IoT Networks and Protocols
- Mobile Learning in Education
- IoT and Edge/Fog Computing
- Non-Invasive Vital Sign Monitoring
- Age of Information Optimization
- Caching and Content Delivery
- AI-based Problem Solving and Planning
- Recommender Systems and Techniques
- Energy Harvesting in Wireless Networks
- Indoor and Outdoor Localization Technologies
- Internet Traffic Analysis and Secure E-voting
- Speech and Audio Processing
- Cryptography and Data Security
- Advanced MIMO Systems Optimization
- Data-Driven Disease Surveillance
- User Authentication and Security Systems
- Ultrasound and Cavitation Phenomena
- Reinforcement Learning in Robotics
- Wireless Body Area Networks
- Bluetooth and Wireless Communication Technologies
- Neural Networks and Applications
- Magnesium Oxide Properties and Applications
- Advanced Neural Network Applications
Macquarie University
2022-2024
Shandong University
2024
Xi'an University of Architecture and Technology
2024
State Key Laboratory of Quantum Optics and Quantum Optics Devices
2023
IMEC
2023
Shanxi University
2023
Lanzhou Jiaotong University
2022
Dalian University of Technology
2018-2022
National University of Defense Technology
2022
UNSW Sydney
2022
Most studies have the achieved rapid and accurate determination of soil organic carbon (SOC) using laboratory spectroscopy; however, it remains difficult to map spatial distribution SOC. To predict SOC at a regional scale, we obtained fourteen hyperspectral images from Gaofen-5 (GF-5) satellite decomposed reconstructed original reflectance (OR) first derivative (FDR) discrete wavelet transform (DWT) different scales. At these scales, as inputs, selected 3 optimal bands with highest weight...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Existing works mainly address the single-domain CTR problem and model aspects such as feature interaction, user behavior history contextual information. Nevertheless, ads are usually displayed with natural content, which offers an opportunity for cross-domain prediction. In this paper, we leverage auxiliary data from source domain to improve performance of target domain. Our study based on UC Toutiao (a...
Federated Learning (FL) is a novel distributed machine learning which allows thousands of edge devices to train model locally without uploading data concentrically the server. But since real federated settings are resource-constrained, FL encountered with systems heterogeneity causes lot stragglers directly and then leads significantly accuracy reduction indirectly. To solve problems caused by heterogeneity, we introduce self-adaptive framework FedSAE adjusts training task automatically...
Federated learning (FL) is an emerging distributed machine paradigm that protects privacy and tackles the problem of isolated data islands. At present, there are two main communication strategies FL: synchronous FL asynchronous FL. The advantages model has high precision fast convergence speed. However, this strategy risk central server waits too long for devices, namely, straggler effect which a negative impact on some time-critical applications. Asynchronous natural advantage in mitigating...
LoRa has emerged as one of the promising long-range and low-power wireless communication technologies for Internet Things (IoT). With massive deployment networks, ability to perform Firmware Update Over-The-Air (FUOTA) is becoming a necessity unattended devices. Alliance recently dedicated specification FUOTA, but existing solution several drawbacks, such low energy efficiency, poor transmission reliability, biased multicast grouping. In this paper, we propose novel energy-efficient,...
In Taobao, the largest e-commerce platform in China, billions of items are provided and typically displayed with their images.For better user experience business effectiveness, Click Through Rate (CTR) prediction online advertising system exploits abundant historical behaviors to identify whether a is interested candidate ad. Enhancing behavior representations images will help understand user's visual preference improve accuracy CTR greatly. So we propose model jointly ID features images....
Facial expression recognition plays a vital role to enable emotional awareness in multimedia Internet of Things applications. Traditional camera or wearable sensor based approaches may compromise user privacy cause discomfort. Recent device-free open promising direction by exploring Wi-Fi ultrasound signals reflected from facial muscle movements, but limitations exist such as poor performance presence body motions and not being able detect multiple targets. To bridge the gap, we propose...
We present FinDroidHR, a novel gesture input technique for off-the-shelf smartwatches. Our is designed to detect 10 hand gestures on the wearing smartwatch. The enabled by analysing features of Photoplethysmography (PPG) signal that optical heart-rate sensors capture. In study with 20 participants, we show FinDroidHR achieves 90.55% accuracy and 90.73% recall. work first explore feasibility using wearable devices recognise gestures. Without requiring bespoke hardware, can be readily used existing
Abstract The existing methods of contact respiration rate (RR) measurement can bring discomfort to the person being measured. However, RR is a human index that has be monitored in clinical medicine. To overcome limitations methods, non-contact method based on an infrared thermal camera proposed. This phenomenon breathing causes periodic temperature changes around nostrils. First, used collect image sequences face. And then, track region-of-interest (ROI) moving sequences, You Only Look Once...
Purpose – A picture is worth a thousand words. Multimedia teaching materials have been widely adopted by teachers in Physics, Biotechnology, Psychology, Religion, Analytical Science, and Economics nowadays. To assist with engaging students their economic study, increase learning efficiency understanding, solve misconception problems, encourage class discussion, final performance for (especially international RA students), some animations cartoons are developed to explain basic concepts both...
Personal mobile sensing is fast permeating our daily lives to enable activity monitoring, healthcare and rehabilitation. Combined with deep learning, these applications have achieved significant success in recent years. Different from conventional cloud-based paradigms, running learning on devices offers several advantages including data privacy preservation low-latency response for both model inference update. Since collection costly reality, Google's Federated Learning not only complete...
The articles in this special section focus on cognitive computing systems. Humans are arguably the most intelligent entities known universe; objective of is to understand and replicate essence human intelligence. Autonomous systems self-contained self-regulated that continuously evolve real time response changes their environment. Fundamental evolution learning development. Cognition basis for autonomous Human cognition refers processes enable humans perform both mundane specialized tasks....
Alzheimer's disease (AD) is a non-treatable and non-reversible that affects about 6% of people who are 65 older. Brain magnetic resonance imaging (MRI) pseudo-3D technology widely used for AD diagnosis. Convolutional neural networks with 3D kernels (3D CNNs) often the default choice deep learning based MRI analysis. However, CNNs usually computationally costly data-hungry. Such disadvantages post barrier using modern techniques in medical domain, which number data can be training limited. In...
Physical-layer key generation is promising in establishing a pair of cryptographic keys for emerging LoRa networks. However, existing systems may perform poorly since the channel reciprocity critically impaired due to low data rate and long range To bridge this gap, paper proposes novel system networks, named ChirpKey. We reveal that underlying limitations are coarse-grained measurement inefficient quantization process. enable fine-grained information, we propose LoRa-specific method...
Radio Frequency (RF)-based gait recognition has emerged as a promising technology to authenticate individuals in pervasive and unobtrusive way. However, fundamental challenge remains collecting extensive data of the same user environment. To address this challenge, paper introduces XGait, cross-modal framework that does not require prior deployment RF devices or explicit collection. The key idea is leverage signals Inertial Measurement Unit (IMU), which widely available modern mobile...
Wireless powered communication technology has a great potential to power low-power wireless sensor networks and Internet of Things (IoT) for real-time applications in future 5G networks, where age information (AoI) plays very important performance metric. This paper studies the system average AoI network, wireless-powered user harvests energy from source (WPS) then transmits data packets its access point (AP) by using harvested energy. The generates with some probability adopts...
Privacy-preserving data publishing has received much attention in recent years. Prior studies have developed various algorithms such as generalization, anatomy, and L-diversity slicing to protect individuals’ privacy when transactional are published for public use. These existing algorithms, however, all certain limitations. For instance, generalization protects identity well but loses a considerable amount of information. Anatomy prevents attribute disclosure lowers information loss, fails...
Blood pressure (BP) measurement is an indispensable tool in diagnosing and treating many diseases such as cardiovascular failure stroke. Traditional direct can be invasive, wearable-based methods may have limitations of discomfort inconvenience. Contact-free BP has been recently advocated a promising alternative. In particular, Millimetre-wave (mmWave) sensing demonstrated its potential, however it confronted with several challenges including noise vulnerability to human's tiny motions which...