119 lines
7.5 KiB
Markdown
119 lines
7.5 KiB
Markdown
# Attacks in FL
|
||
Here we present recent works on the attacks in FL.
|
||
|
||
|[Survey](#survey)|[Privacy Attacks](#privacy-attacks-in-fl)|[Backdoor Attacks](#backdoor-attacks-in-fl)|[Untargeted Attacks](#untargeted-attacks-in-fl)|
|
||
|
||
|
||
## Survey
|
||
| Title | Venue | Link | Year
|
||
| ------------------------------------------------------------ | ---------- |---------------------------------------------|-----------|
|
||
| A Survey on Gradient Inversion: Attacks, Defenses and Future Directions | arxiv | [pdf](https://arxiv.org/pdf/2206.07284.pdf) | 2022 |
|
||
| Threats to Federated Learning: A Survey | arxiv| [pdf](https://arxiv.org/pdf/2003.02133.pdf) | 2020 |
|
||
|
||
|
||
## Privacy Attacks in FL
|
||
|
||
## 2022
|
||
| Title | Venue | Link |
|
||
| ------------------------------------------------------------ | ---------- |---------------------------------------------|
|
||
| Robbing the Fed: Directly Obtaining Private Data in Federated Learning with Modified Models| ICLR | [pdf](https://openreview.net/pdf?id=fwzUgo0FM9v) |
|
||
|Fishing for User Data in Large-Batch Federated Learning via Gradient Magnification |Arxiv | [pdf](https://arxiv.org/pdf/2202.00580.pdf)|
|
||
|Bayesian Framework for Gradient Leakage|[pdf](https://openreview.net/pdf?id=f2lrIbGx3x7)|
|
||
|Auditing Privacy Defenses in Federated Learning via Generative Gradient Leakage|CVPR|[pdf](https://openaccess.thecvf.com/content/CVPR2022/papers/Li_Auditing_Privacy_Defenses_in_Federated_Learning_via_Generative_Gradient_Leakage_CVPR_2022_paper.pdf)|
|
||
|
||
|
||
## 2021
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
| See through gradients: Image batch recovery via gradinversion | CVPR | [pdf](https://openaccess.thecvf.com/content/CVPR2021/papers/Yin_See_Through_Gradients_Image_Batch_Recovery_via_GradInversion_CVPR_2021_paper.pdf) |
|
||
| Gradient disaggregation: Breaking privacy in federated learning by reconstructing the user participant matrix | ICML | [pdf](http://proceedings.mlr.press/v139/lam21b/lam21b.pdf) |
|
||
| Evaluating gradient inversion attacks and defenses in federated learning|NeurIPS|[pdf](https://proceedings.neurips.cc/paper/2021/file/3b3fff6463464959dcd1b68d0320f781-Paper.pdf)|
|
||
| Bayesian framework for gradient leakage|ICLR|[pdf](https://arxiv.org/pdf/2111.04706.pdf)|
|
||
| Catastrophic data leakage in vertical federated learning. |NeurIPS|[pdf](https://proceedings.neurips.cc/paper/2021/file/08040837089cdf46631a10aca5258e16-Paper.pdf)|
|
||
| Gradient inversion with generative image prior|NeurIPS|[pdf](https://proceedings.neurips.cc/paper/2021/file/fa84632d742f2729dc32ce8cb5d49733-Paper.pdf)|
|
||
| R-gap: Recursive gradient attack on privacy|ICLR|[pdf](https://openreview.net/pdf?id=RSU17UoKfJF)|
|
||
| Understanding training-data leakage from gradients in neural networks for image classifications. |NeurIPS workshop|[pdf](https://arxiv.org/pdf/2111.10178.pdf)|
|
||
|Soteria: Provable Defense against Privacy Leakage in Federated Learning from Representation Perspective|CVPR|[pdf](https://openaccess.thecvf.com/content/CVPR2021/papers/Sun_Soteria_Provable_Defense_Against_Privacy_Leakage_in_Federated_Learning_From_CVPR_2021_paper.pdf)|
|
||
|Reconstruction Attack on Instance Encoding for Language Understanding|EMNLP|[pdf](https://aclanthology.org/2021.emnlp-main.154.pdf)|
|
||
|Source Inference Attacks in Federated Learning|ICDM|[pdf](https://arxiv.org/pdf/2109.05659.pdf)|
|
||
|TAG: Gradient Attack on Transformer-based Language Models|EMNLP(findings)|[pdf](https://aclanthology.org/2021.findings-emnlp.305.pdf)|
|
||
|Unleashing the Tiger: Inference Attacks on Split Learning|CCS|[pdf](https://dl.acm.org/doi/pdf/10.1145/3460120.3485259)|
|
||
|
||
|
||
## 2020
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
| idlg: Improved deep leakage from gradients | arxiv | [pdf](https://arxiv.org/pdf/2001.02610.pdf) |
|
||
| A framework for evaluating client privacy leakages in federated learning | ESORICS | [pdf](https://arxiv.org/pdf/2004.10397.pdf) |
|
||
| Inverting gradients – how easy is it to break privacy in federated learning? |NeurIPS| [pdf](https://proceedings.neurips.cc/paper/2020/file/c4ede56bbd98819ae6112b20ac6bf145-Paper.pdf)|
|
||
| Sapag: A self adaptive privacy attack from gradients|arxiv |[pdf](https://arxiv.org/pdf/2009.06228.pdf)|
|
||
| Is Private Learning Possible with Instance Encoding?|S&P|[pdf](https://arxiv.org/pdf/2011.05315.pdf)|
|
||
|
||
|
||
## 2019
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
| Deep Leakage from Gradients | NeurIPS | [pdf](https://papers.nips.cc/paper/2019/file/60a6c4002cc7b29142def8871531281a-Paper.pdf) |
|
||
| Beyond Inferring Class Representatives: User-Level Privacy Leakage From Federated Learning|Infocom|[pdf](https://arxiv.org/pdf/1812.00535.pdf)|
|
||
| Exploiting Unintended Feature Leakage in Collaborative Learning|S&P|[pdf](https://arxiv.org/pdf/1805.04049.pdf)|
|
||
| Comprehensive Privacy Analysis of Deep Learning: Passive and Active White-box Inference Attacks against Centralized and Federated Learning|S&P|[pdf](https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8835245)|
|
||
|
||
|
||
|
||
## 2017
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
| Deep Models Under the GAN: Information Leakage from Collaborative Deep Learning | CCS | [pdf](https://arxiv.org/pdf/1702.07464.pdf) |
|
||
|
||
|
||
## Backdoor Attacks in FL
|
||
## 2022
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Narcissus: A Practical Clean-Label Backdoor Attack with Limited Information |arxiv|[pdf](https://arxiv.org/pdf/2204.05255.pdf)|
|
||
|Neurotoxin: Durable Backdoors in Federated Learning|arxiv|[pdf](https://www2.eecs.berkeley.edu/Pubs/TechRpts/2022/EECS-2022-89.pdf)|
|
||
## 2021
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|WaNet - Imperceptible Warping-based Backdoor Attack |ICLR|[pdf](https://arxiv.org/pdf/2102.10369.pdf)|
|
||
|
||
## 2020
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Attack of the Tails: Yes, You Really Can Backdoor Federated Learning|NeurIPS|[pdf](https://papers.nips.cc/paper/2020/file/b8ffa41d4e492f0fad2f13e29e1762eb-Paper.pdf)|
|
||
|DBA: Distributed Backdoor Attacks against Federated Learning|ICLR|[pdf](https://openreview.net/pdf?id=rkgyS0VFvr)|
|
||
|How To Backdoor Federated Learning|AISTATS|[pdf](https://arxiv.org/pdf/1807.00459.pdf)|
|
||
|
||
## 2019
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|BadNets: Evaluating Backdooring Attacks on Deep Neural Networks|IEEE Access|[pdf](https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8685687)|
|
||
|Analyzing Federated Learning through an Adversarial Lens|NeurIPS|[pdf](https://arxiv.org/pdf/1811.12470.pdf)|
|
||
|
||
## 2017
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning |arxiv|[pdf](https://arxiv.org/pdf/1712.05526.pdf)|
|
||
|
||
|
||
## Untargeted Attacks in FL
|
||
## 2022
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Learning to Attack Federated Learning: A Model-based Reinforcement Learning Attack Framework|NeurIPS|[pdf](https://openreview.net/pdf?id=4OHRr7gmhd4)|
|
||
|Poisoning Deep Learning Based Recommender Model in Federated Learning Scenarios | IJCAI |[pdf](https://www.ijcai.org/proceedings/2022/0306.pdf)
|
||
|
||
|
||
## 2021
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Manipulating the byzantine: Optimizing model poisoning attacks and defenses for federated learning|NDSS|[pdf](https://par.nsf.gov/servlets/purl/10286354)|
|
||
|
||
## 2020
|
||
| Title | Venue | Link |
|
||
| --- | --- | --- |
|
||
|Local Model Poisoning Attacks to Byzantine-Robust Federated Learning|USENIX SCEURITY|[pdf](https://www.usenix.org/system/files/sec20summer_fang_prepub.pdf)|
|
||
|Fall of Empires: Breaking Byzantine-tolerant SGD by Inner Product Manipulation|UAI|[pdf](http://proceedings.mlr.press/v115/xie20a/xie20a.pdf)|
|
||
|
||
|