Gradient inversion-based inference attack against federated learning
Federated learning is a state-of-the-art paradigm where deep learning models based on servers can be trained without having direct access to private training data. In federated learning, clients transfer gradients to the server, which can be used to further improve the model. However, the gradients...
| Автор: | |
|---|---|
| Інші автори: | |
| Формат: | Final Year Project (FYP) |
| Мова: | English |
| Опубліковано: |
Nanyang Technological University
2023
|
| Предмети: | |
| Онлайн доступ: | https://hdl.handle.net/10356/172760 |
| _version_ | 1826129163104288768 |
|---|---|
| author | Chan, Joel Yuan Wei |
| author2 | Chang Chip Hong |
| author_facet | Chang Chip Hong Chan, Joel Yuan Wei |
| author_sort | Chan, Joel Yuan Wei |
| collection | NTU |
| description | Federated learning is a state-of-the-art paradigm where deep learning models based on servers can be trained without having direct access to private training data. In federated learning, clients transfer gradients to the server, which can be used to further improve the model. However, the gradients transferred are susceptible to leaking the private data to the server, and this is a concern in many real-life applications, such as medical image classification. This attack is called gradient inversion.
In this project, a specific gradient inversion attack, using generative adversarial networks to generate an image prior, will be implemented on a simulated federated learning paradigm. By obtaining the gradients, this project will demonstrate how human facial images can be reconstructed simply from those gradients, thereby showing that federated learning is not a privacy-preserving paradigm. Analysis of the experimental data also shows that increasing the batch size or the image dimensions can affect the quality of the reconstructed images. Lastly, some suggestions on future work pertaining to implementation of federated learning in language models, along with gradient inversion defense techniques, are discussed. |
| first_indexed | 2024-10-01T07:36:31Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/172760 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T07:36:31Z |
| publishDate | 2023 |
| publisher | Nanyang Technological University |
| record_format | dspace |
| spelling | ntu-10356/1727602023-12-22T15:43:40Z Gradient inversion-based inference attack against federated learning Chan, Joel Yuan Wei Chang Chip Hong School of Electrical and Electronic Engineering ECHChang@ntu.edu.sg Engineering::Electrical and electronic engineering Federated learning is a state-of-the-art paradigm where deep learning models based on servers can be trained without having direct access to private training data. In federated learning, clients transfer gradients to the server, which can be used to further improve the model. However, the gradients transferred are susceptible to leaking the private data to the server, and this is a concern in many real-life applications, such as medical image classification. This attack is called gradient inversion. In this project, a specific gradient inversion attack, using generative adversarial networks to generate an image prior, will be implemented on a simulated federated learning paradigm. By obtaining the gradients, this project will demonstrate how human facial images can be reconstructed simply from those gradients, thereby showing that federated learning is not a privacy-preserving paradigm. Analysis of the experimental data also shows that increasing the batch size or the image dimensions can affect the quality of the reconstructed images. Lastly, some suggestions on future work pertaining to implementation of federated learning in language models, along with gradient inversion defense techniques, are discussed. Bachelor of Engineering (Information Engineering and Media) 2023-12-19T23:51:40Z 2023-12-19T23:51:40Z 2023 Final Year Project (FYP) Chan, J. Y. W. (2023). Gradient inversion-based inference attack against federated learning. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/172760 https://hdl.handle.net/10356/172760 en A2308-222 application/pdf Nanyang Technological University |
| spellingShingle | Engineering::Electrical and electronic engineering Chan, Joel Yuan Wei Gradient inversion-based inference attack against federated learning |
| title | Gradient inversion-based inference attack against federated learning |
| title_full | Gradient inversion-based inference attack against federated learning |
| title_fullStr | Gradient inversion-based inference attack against federated learning |
| title_full_unstemmed | Gradient inversion-based inference attack against federated learning |
| title_short | Gradient inversion-based inference attack against federated learning |
| title_sort | gradient inversion based inference attack against federated learning |
| topic | Engineering::Electrical and electronic engineering |
| url | https://hdl.handle.net/10356/172760 |
| work_keys_str_mv | AT chanjoelyuanwei gradientinversionbasedinferenceattackagainstfederatedlearning |