Search

Your search keyword '"Ge, Jingyi"' showing total 26 results
Start Over Author "Ge, Jingyi"
26 results on '"Ge, Jingyi"'

3. White-box Inference Attacks against Centralized Machine Learning and Federated Learning

Author

Ge, Jingyi

Subjects
Computer Science - Cryptography and Security, Computer Science - Machine Learning
Abstract

With the development of information science and technology, various industries have generated massive amounts of data, and machine learning is widely used in the analysis of big data. However, if the privacy of machine learning applications' customers cannot be guaranteed, it will cause security threats and losses to users' personal privacy information and service providers. Therefore, the issue of privacy protection of machine learning has received wide attention. For centralized machine learning models, we evaluate the impact of different neural network layers, gradient, gradient norm, and fine-tuned models on member inference attack performance with prior knowledge; For the federated learning model, we discuss the location of the attacker in the target model and its attack mode. The results show that the centralized machine learning model shows more serious member information leakage in all aspects, and the accuracy of the attacker in the central parameter server is significantly higher than the local Inference attacks as participants., Comment: 17 papers, 3 figures, 6 tables

Published
2022

7. Strategies to improve the therapeutic efficacy of mesenchymal stem cell‐derived extracellular vesicle (MSC-EV): a promising cell-free therapy for liver disease

Author

Zheng, Lijuan, primary, Gong, Hui, additional, Zhang, Jing, additional, Guo, Linna, additional, Zhai, Zhuofan, additional, Xia, Shuang, additional, Hu, Zhiyu, additional, Chang, Jing, additional, Jiang, Yizhu, additional, Huang, Xinran, additional, Ge, Jingyi, additional, Zhang, Bikui, additional, and Yan, Miao, additional

Published
2023
Full Text
View/download PDF

14. White-box Inference Attacks against Centralized Machine Learning and Federated Learning

Author

Ge, Jingyi

Subjects
FOS: Computer and information sciences, Computer Science - Machine Learning, Computer Science - Cryptography and Security, Cryptography and Security (cs.CR), Machine Learning (cs.LG)
Abstract

With the development of information science and technology, various industries have generated massive amounts of data, and machine learning is widely used in the analysis of big data. However, if the privacy of machine learning applications' customers cannot be guaranteed, it will cause security threats and losses to users' personal privacy information and service providers. Therefore, the issue of privacy protection of machine learning has received wide attention. For centralized machine learning models, we evaluate the impact of different neural network layers, gradient, gradient norm, and fine-tuned models on member inference attack performance with prior knowledge; For the federated learning model, we discuss the location of the attacker in the target model and its attack mode. The results show that the centralized machine learning model shows more serious member information leakage in all aspects, and the accuracy of the attacker in the central parameter server is significantly higher than the local Inference attacks as participants., Comment: 17 papers, 3 figures, 6 tables

Published
2023
Full Text
View/download PDF

21. Path Planning for Complex Thin-Walled Structures in 3D Printing: Improved Q-Learning Method.

Author

WANG Yi, GE Jingyi, XUE Xinwei, WANG Shengfa, and LI Fengqi

Subjects
THIN-walled structures, THREE-dimensional printing, CONSTRAINT algorithms, REINFORCEMENT learning
Abstract

3D printing is an advanced manufacturing technology, the optimized path planning method of 3D printing can improve efficiency or molding quality. For the traditional path planning methods are short of handling complex thinwalled structures, an reinforcement learning-based path planning method named Q-path is proposed for complex thin-walled structures. First of all, based on the path planning in 3D printing is a filling task, the path planning task in reinforcement learning is converted to a traversal problem. In order to improve printing efficiency and molding quality, the optimization goal is to minimize the total cost of printing, and the constraint conditions in the reinforcement learning are designed according to the optimization goal, that is, to minimize the number of start, stop and turns of the print head. This paper establishes a simulation environment of a slice, uses the Q-learning algorithm with the above constraints, learning is guided by calculating the value of the total cost to find the optimal path plan. Experimental results show that this method is better than the traditional method for 3D printing path planning in printing complex thin-walled structures. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

22. Path Planning for Complex Thin-Walled Structures in 3D Printing:Improved Q-Learning Method.

Author

WANG Yi, GE Jingyi, XUE Xinwei, WANG Shengfa, and LI Fengqi

Subjects
THIN-walled structures, THREE-dimensional printing, CONSTRAINT algorithms, REINFORCEMENT learning
Abstract

3D printing is an advanced manufacturing technology, the optimized path planning method of 3D printing can improve efficiency or molding quality. For the traditional path planning methods are short of handling complex thin-walled structures, an reinforcement learning-based path planning method named Q-path is proposed for complex thin-walled structures. First of all, based on the path planning in 3D printing is a filling task, the path planning task in reinforcement learning is converted to a traversal problem. In order to improve printing efficiency and molding quality, the optimization goal is to minimize the total cost of printing, and the constraint conditions in the reinforcement learning are designed according to the optimization goal, that is, to minimize the number of start, stop and turns of the print head. This paper establishes a simulation environment of a slice, uses the Q-learning algorithm with the above constraints, learning is guided by calculating the value of the total cost to find the optimal path plan. Experimental results show that this method is better than the traditional method for 3D printing path planning in printing complex thin-walled structures. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results