A5018927991- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- 3D Shape Modeling and Analysis
- Text and Document Classification Technologies
- Face recognition and analysis
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Industrial Vision Systems and Defect Detection
- Advanced Data Storage Technologies
- Physical Unclonable Functions (PUFs) and Hardware Security
- Caching and Content Delivery
- Advanced Malware Detection Techniques
- Medical Image Segmentation Techniques
- Organic Light-Emitting Diodes Research
- Metaheuristic Optimization Algorithms Research
- Diabetic Foot Ulcer Assessment and Management
- Spam and Phishing Detection
- Web Data Mining and Analysis
- Advanced Vision and Imaging
- Face and Expression Recognition
- Computer Graphics and Visualization Techniques
- 3D Surveying and Cultural Heritage
- Parallel Computing and Optimization Techniques
Xi'an University of Technology
2018-2025
University of Electronic Science and Technology of China
2013-2025
University of Chinese Academy of Sciences
2024
Chinese Academy of Sciences
2024
Institute of Microelectronics
2024
Ping An (China)
2024
Qingdao University of Technology
2022
Shanghai Jiao Tong University
2019-2022
Qingdao University of Science and Technology
2022
Chongqing University
2019-2021
Three-dimensional object detection based on deep neural networks (DNNs) is widely used in safety-related applications, such as autonomous driving. However, existing research has shown that 3D models are vulnerable to adversarial attacks. Hence, the improvement robustness of under attacks investigated this work. A autoencoder-based anomaly method proposed, which a strong ability detect elaborate samples an unsupervised way. The proposed operates given Light Detection and Ranging (LiDAR) scene...
We report a hydrophobically associating hydrogel. This conductive hydrogel boasts excellent stretchability, outstanding anti-swelling ability, and strong underwater adhesion. The hydrogel-based sensor holds great promise for communication.
An anthracene carboxamide derivative of the excited-state intramolecular proton-transfer compound 2-(2'-hydroxyphenyl)benzothiazole has been newly developed to produce prominent characteristics aggregation-induced enhanced emission (AIEE) with a high solid-state fluorescence quantum efficiency 78.1%. Compared our previously reported phenyl derivatives, small tailoring molecular structure was found result in big difference dominant factor AIEE mechanism. In mechanism is restriction twisted...
This study developed a new solid-state, highly emissive phenylbenzoxazole-based organic compound, N-(4-(benzo[d]oxazol-2-yl)phenyl)-4-tert-butylbenzamide, which exhibited distinct aggregation-induced enhanced emission. The solid fluorescence efficiency of the newly compound was 50.3%, whereas that in THF solution only 0.22%. single-crystal analyses revealed specific three-dimensional #-shaped cross stacking between molecules observed solid/aggregated state, driven by C–H···π interaction and...
Spherical materials with yolk-shell structure have great potential for a wide range of applications. The main advantage the geometry is possibility introducing different chemical or physical properties within single particle. Here, one-step hydrothermal synthesis route fabricating amphoteric structured aluminum phenylphosphonate microspheres using urea as precipitant proposed. resulting display 3D sphere-in-sphere architecture anionic core and cationic shell. controllable phosphates various...
In this paper, the target node localization problems based on hybrid RSS-AOA measurements in both noncooperative and cooperative three-dimensional (3-D) wireless sensor networks (WSNs) are discussed. By using novel error approximate expressions for received signal strength (RSS) angle-of-arrival (AOA) measurement models, new estimators least squares (LS) criterion proposed. These can be transformed into mixed semi-definite programming (SDP) second-order cone (SOCP) by applying convex...
As an emerging threat to deep neural networks (DNNs), backdoor attacks have received increasing attentions due the challenges posed by lack of transparency inherent in DNNs. In this article, we develop efficient algorithm from interpretability DNNs defend against DNN models. To extract critical neurons, deploy sets control gates following neurons layers, and function a model can be interpreted as semantic sensitivities input samples. A identification approach, derived activation frequency...
Head pose estimation from depth image is a challenging problem, considering its large variations, severer occlusions, and low quality of data. In contrast to existing approaches that take 2D as input, we propose novel deep regression architecture called PointNet, which consumes 3D point sets derived describing the visible surface head. To cope with non-stationary property variation process, network facilitated an ordinal module incorporates metric penalties into ground truth label...
Distress is a continuous process of unpleasant experience, which could be normal emotional response sadness, fear, and fatigue, but it may also deteriorate into depression, anxiety, panic, other mental crises without early diagnosis intervention (Riba et al., 2019). Many factors, such as psychological, social, spiritual environment, are involved in the pathogenesis distress (Alfonsson 2016). common patients with cancer, was estimated Zabora al.
Abstract 3D object tracking based on deep neural networks has a wide range of potential applications, such as autonomous driving and robotics. However, are vulnerable to adversarial examples. Traditionally, examples generated by applying perturbations individual samples, which requires exhaustive calculations for each sample thereby suffers from low efficiency during malicious attacks. Hence, the universal perturbation been introduced, is sample-agnostic. The able make classifiers...
Existing extended one-versus-rest multi-label support vector machine (OVR-ESVM) adopting non-linear kernel is seriously restricted by excessive training time when it applied to large-scale data set. In order overcome this problem, we improve the OVR-ESVM introducing principle of approximate extreme points and new ranking loss construct a novel using (AEML-ESVM). By optimizing only on representative set which can be acquired via method, AEML-ESVM classification algorithm substantially shorten...
Geometric analysis of three-dimensional (3D) surfaces with local deformations is a challenging task, required by mobile devices. In this paper, we propose new feature-based method derived from diffusion geometry, including keypoint detector named persistence-based Heat Kernel Signature (pHKS), and feature descriptor Propagation Strips (HeaPS). The pHKS first constructs scalar field using the heat kernel signature function. generated at small scale to capture fine geometric information...