A5110097903- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Adversarial Robustness in Machine Learning
- Industrial Technology and Control Systems
- Network Security and Intrusion Detection
- Traffic Prediction and Management Techniques
- Stochastic Gradient Optimization Techniques
- Fatigue and fracture mechanics
- Information and Cyber Security
- Software Testing and Debugging Techniques
- Risk and Safety Analysis
- Advanced Wireless Communication Technologies
- Vehicle License Plate Recognition
- Anomaly Detection Techniques and Applications
- Digital Marketing and Social Media
- Cooperative Communication and Network Coding
- Railway Engineering and Dynamics
- E-commerce and Technology Innovations
- Technology and Security Systems
- Statistical and Computational Modeling
- Full-Duplex Wireless Communications
- Advanced Graph Neural Networks
- Advanced Sensor and Control Systems
- Advanced Computational Techniques and Applications
- Simulation and Modeling Applications
Shanghai International Studies University
2024
Rensselaer Polytechnic Institute
2021-2024
China Aerospace Science and Technology Corporation
2023
China Jiliang University
2014
China Agricultural University
2013
Horizontal Federated learning (FL) handles multi-client data that share the same set of features, and vertical FL trains a better predictor combine all features from different clients. This paper targets solving in an asynchronous fashion, develops simple method. The new method allows each client to run stochastic gradient algorithms without coordination with other clients, so it is suitable for intermittent connectivity further uses technique perturbed local embedding ensure privacy improve...
Recent studies show that private training data can be leaked through the gradients sharing mechanism deployed in distributed machine learning systems, such as federated (FL). Increasing batch size to complicate recovery is often viewed a promising defense strategy against leakage. In this paper, we revisit premise and propose an advanced leakage attack with theoretical justification efficiently recover from shared aggregated gradients. We name our proposed method catastrophic vertical...
Established in Nanjing 2012, SHEIN has emerged as a powerhouse the global e-commerce realm, boasting sales exceeding 100 billion yuan by 2021 and dethroning Amazon most downloaded shopping app. This paper examines how SHEIN, mastering people's psychology, employs an integrated marketing strategy that combines social media, user-generated content, search engine optimization, short videos, live streaming to get consumers interested its clothing [4]. The analysis is made based on Demand Theory...
Considering the actual requirements of data and process management in agricultural machinery product design, a Windchill-based PDM system is proposed for chassis products design regarding its architecture, models functional modules. Specific workflow are built to provide collaborative platform management. It consists seven major modules such as management, search, team, version permission integration, etc. Its implementation methods will useful guide application manufacturing enterprises...
Abstract With the continuous increasing of service-rail-vehicle in China, some defects such as cracks appeared at different stages service when they had not reached design life, while current theories and methods cannot explain appearance defects, which brought risks hidden dangers to operation vehicles. It was found that reasons were mainly caused by one or several factors material performances, design, manufacturing process operating environment. Since more sensors being installed on...
In this paper, a new deep reinforcement learning algorithm DEU-DRQNper is proposed for penetration test path planning under partially observable conditions. The improved on the basis of DRQN algorithm, including experience playback priority sequence, action exploration and selection strategy combining ɛ greedy UCB way Double DQN calculating target Q value. Priority sequence mechanism selects more valuable sequences according to probability distribution corresponding during sampling...
in this paper, the key technology of traffic statement automatic detection is researched, based on analyzing properties information from collected license plates, a new detective method pointed out, can express essential characters statement, and valuably gain precise degree. This expresses fine applicative prospect field real time management.
In this paper, a novel cooperation space-time network coding (CSTNC) scheme based on multi-sources distributed cooperative is considered. scheme, multiple sources transmit signals to relay and destination node. The re-encodes the decoded through with aid of antennas. simulation shows that BER, capacity outage probability CSTNC becomes better increasing antenna number very close traditional (TCSTC) scheme. Meanwhile, transmission efficiency analysis demonstrates has higher rate.
This article has been retracted by the publisher.
The paper presents a structure and functions of an expert system for aided design rice harvester chassis systems. It was developed on basis detailed analysis the process, including knowledge bases regarding systems automated design, community parameterized parts assemblies interference engine. In creation system, Visual C++, Pro/Engineering Oracle database were used. This is characterized by rule-oriented representation knowledge, forward chaining inference methods, graphic results...
Based on the change of distance function and mobile direction angle which are from vehicle GPS systems, a novel traffic incident automatic detection system is designed in. Not only does principle this method deduced, but properties experiment analyzed. With experiment, detective expresses some advantage properties.