A5100442739- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Advanced Malware Detection Techniques
- Network Security and Intrusion Detection
- Internet Traffic Analysis and Secure E-voting
- Digital Media Forensic Detection
- Advanced Neural Network Applications
- Blockchain Technology Applications and Security
- Robotics and Sensor-Based Localization
- Robotic Path Planning Algorithms
- Advancements in Photolithography Techniques
- Advanced Algorithms and Applications
- Face and Expression Recognition
- Blind Source Separation Techniques
- Advanced Sensor and Control Systems
- Video Surveillance and Tracking Methods
- Biometric Identification and Security
- Energy Efficient Wireless Sensor Networks
- Wireless Signal Modulation Classification
- Full-Duplex Wireless Communications
- User Authentication and Security Systems
- Advanced Surface Polishing Techniques
- Aerospace Engineering and Control Systems
- Autonomous Vehicle Technology and Safety
- Emotion and Mood Recognition
China Academy of Engineering Physics
2020-2024
Chinese Academy of Engineering
2024
Tianjin University
2021-2023
Zhongji Test Equipment (China)
2023
Tongji University
2023
China Academy of Urban Planning and Design
2023
Suzhou Research Institute
2023
University of Electronic Science and Technology of China
2016-2022
State Key Laboratory of Crystal Materials
2022
Cloud Computing Center
2022
Many IoT (Internet of Things) systems run Android or Android-like systems. With the continuous development machine learning algorithms, learning-based malware detection system for devices has gradually increased. However, these models are often vulnerable to adversarial samples. An automated testing framework is needed help perform security analysis. The current methods generating samples mostly require training parameters and most aimed at image data. To solve this problem, we propose a...
In the current intranet environment, information is becoming more readily accessed and replicated across a wide range of interconnected systems. Anyone using computer may access content that he does not have permission to access. For an insider attacker, it relatively easy steal colleague’s password or use unattended launch attack. A common one-time user authentication method work in this situation. paper, we propose based on mouse biobehavioral characteristics deep learning, which can...
Model inversion attacks (MIAs) aim to recover private data from inaccessible training sets of deep learning models, posing a privacy threat. MIAs primarily focus on the white-box scenario where attackers have full access model's structure and parameters. However, practical applications are usually in black-box scenarios or label-only scenarios, i.e., can only obtain output confidence vectors labels by accessing model. Therefore, attack models existing difficult effectively train with...
Recent studies have highlighted audio adversarial examples as a ubiquitous threat to state-of-the-art automatic speech recognition systems. Thorough on how effectively generate are essential prevent potential attacks. Despite many research this, the efficiency and robustness of existing works not yet satisfactory. In this paper, we propose weighted-sampling examples, focusing numbers weights distortion reinforce attack. Further, apply denoising method in loss function make attack more...
In terms of intelligent driving, the adversarial example an attack against traffic signs will cause vehicle to make wrong judgments and decisions. However, existing examples defense algorithms generally have problems such as high training costs poor effects struggle adapt environment driving. order reduce cost while improving accuracy classification, we propose a novel algorithm for combining micro-network structure generative network (GAN). The compresses classification model Discriminator....
Abstract Pointer instruments, widely used in complex environments, present significant challenges for manual reading due to inefficiency and difficulty. For intelligent devices, problems arise with instrument localization, low inference accuracy, incorrect readings. To address these issues, this paper proposes a novel deep-learning algorithm that divides the task of pointer instruments into three stages. First, NV-YOLOv8 model is introduced identify instrument’s location, Spatial Transformer...
Automatic driving has become a extremely hot issue in recent years, and the detection of stop signs is critical for autonomous driving. Different from precious methods which target features were extracted then feed to SVM classifier classify different types traffic signs, this paper introduces kind transfer learning method based on convolutional neural network(CNN). A deep convolution network trained using large data sets, valid region network(RCNN) can be obtained through small amount...
Early and highly precise detection is essential for delaying the progression of coronary artery disease (CAD). Previous methods primarily based on single-modal data inherently lack sufficient information that compromises precision. This paper proposes a novel multi-modal learning method aimed to enhance CAD by integrating ECG, PCG, coupling signals. A signal initially generated operating deconvolution ECG PCG. Then, various entropy features are extracted from its signals, as well recurrence...
The insider threats have always been one of the most severe challenges to cybersecurity. It can lead destruction organisation’s internal network system and information leakage, which seriously threaten confidentiality, integrity availability data. To make matters worse, since attacker has authorized access network, they launch attack from inside erase their trace, makes it challenging track forensics. A blockchain traceability for is proposed in this paper mitigate issue. First, constructs...
With the popularization of IoT (Internet Things) devices and continuous development machine learning algorithms, learning-based malicious traffic detection technologies have gradually matured. However, models are usually very vulnerable to adversarial samples. There is a great need for an automated testing framework help security analysts detect errors in systems. At present, most methods generating samples require training parameters known only applicable image data. To address challenge,...
With the increasing popularity of Unmanned Aerial Vehicles (UAVs), accuracy detecting small objects in large-view images is also expected to increase. However, accurate object detection still a challenging problem. Currently, Image Pyramid Network, Feature Network (FPN), rich training strategies and data augmentation are widely used address this To accurately detect objects, most important thing mine for more feature information. We propose Widened Residual Block (WRB) break through...
In this paper, the cycle performance and capacity fading of commercial C/LiFePO4 batteries were studied at different rate. The electrochemical impedance spectroscopy (EIS) battery was tested analyzed in charge-discharge cycles. morphologies anode cathode numbers also characterized by scanning electron microscope (SEM). experimental results show that become faster higher rate cycling. SEM images cycled electrodes exhibit is possibly attributed to degradation structure. When deteriorates...
For face recognition systems, impostors can obtain legal identity authentication by presenting the printed images, downloaded images or candid videos to sensor. In this paper, an enhanced local binary feature (ELBP) of a map is extracted as classification identify whether real fake face. Compared with dynamic static methods proposed in previous literature, method convenient and effective. We achieved over 95% correct rate on NUAA datasets. Through comprehensive analysis comparison other...
Advanced cyberattacks are often featured by multiple types, layers, and stages, with the goal of cheating monitors. Existing anomaly detection systems usually search logs or traffics alone for evidence attacks but ignore further analysis about attack processes. For instance, traffic methods can only detect flows roughly fail to reconstruct event process reveal current network node status. As a result, they cannot fully model complex multistage attack. To address these problems, we present...
It will be dangerous if hazard gas or inflammable arises in room without being noticed for a long time. A robot is good choice to automatically detect the and send out accurate information alarm. To obtain exact plentiful danger information, has search source. Traditional source searching ways not satisfying. An intelligent mounted sensors, laser finder vision introduced, responding recover delay features of sensors analysed. The requirement feature indoor discussed. two kinds strategy...