Federated Learning

人工智能(Artificial Intelligence, AI)进入以深度学习为主导的大数据时代,基于大数据的机器学习既推动了AI的蓬勃发展,也带来了一系列安全隐患。这些隐患来源于深度学习本身的学习机制,无论是在它的模型建造(训练)阶段,还是在模型推理和使用阶段。这些安全隐患如果被有意或无意地滥用,后果将十分严重。


联邦学习是一种 隐私保护、数据本地存储与计算 的机器学习算法。

文献参考

Part 1: Introduction

Part 2: Survey

Part 3: Benchmarks

Part 4: Model Aggregation

Part 5: Statistical Heterogeneity

5.1 Meta Learning

5.2 Multi-task Learning

5.3 Convergence

5.4 Hierarchical FL

5.5 Transfer Learning

5.6 Graph Neural Network(GNN)

Part 6: System

6.1 Neural Architecture Search

Part 7: Communication Efficiency

7.1 Compression

7.2 Important-Based Updating

7.3 Decentralization

Part 8: Resource Allocation

8.1 Participants Selection

8.2 Adaptive Aggregation

8.3 Incentive Mechanism

Part 9: Vertical Federated Learning

Part 10: Adversarial Attacks

10.1 Wireless Communication and Cloud Computing

Part 11: Data Privacy and Confidentiality

11.1 Courses

11.2 Differential Privacy

11.3 Secure Multi-party Computation

Secret Sharing

Build Safe AI Series

MPC related Paper

Helen: Maliciously Secure Coopetitive Learning for Linear Models (NIPS 2019 Workshop)

11.4 Privacy Preserving Machine Learning

Part 12: Applications

12.1 Healthcare

12.2 Natual Language Processing

Google

Snips

12.3 Computer Vision

12.4 Recommendation

12.5 Industrial

Part 13: Organizations and Companies

13.1 国内篇

微众银行开源 FATE 框架.

Qiang Yang, Tianjian Chen, Yang Liu, Yongxin Tong.

字节跳动开源 FedLearner 框架.

Jiankai Sun, Weihao Gao, Hongyi Zhang, Junyuan Xie.《Label Leakage and Protection in Two-party Split learning》

华控清交 PrivPy 多方计算平台

Yi Li, Wei Xu.《PrivPy: General and Scalable Privacy-Preserving Data Mining》

同盾科技 同盾志邦知识联邦平台

Hongyu Li, Dan Meng, Hong Wang, Xiaolin Li.

百度 MesaTEE 安全计算平台

Tongxin Li, Yu Ding, Yulong Zhang, Tao Wei.《gbdt-rs: Fast and Trustworthy Gradient Boosting Decision Tree》

矩阵元 Rosetta 隐私开源框架
百度 PaddlePaddle 开源联邦学习框架
蚂蚁区块链科技 蚂蚁链摩斯安全计算平台
阿里云 DataTrust 隐私增强计算平台
百度百度点石联邦学习平台
富数科技 阿凡达安全计算平台
香港理工大学

《FedVision: An Online Visual Object Detection Platform Powered by Federated Learning》

《BatchCrypt: Efficient Homomorphic Encryption for Cross-Silo Federated Learning》

《Abnormal Client Behavior Detection in Federated Learning》

北京航空航天大学

《Federated machine learning: Concept and applications》

《Failure Prediction in Production Line Based on Federated Learning: An Empirical Study》

13.2 国际篇

Google 提出 Federated Learning. H. Brendan McMahan. Daniel Ramage. Jakub Konečný. Kallista A. Bonawitz. Hubert Eichner.

《Communication-efficient learning of deep networks from decentralized data》

《Federated Learning: Strategies for Improving Communication Efficiency》

《Advances and Open Problems in Federated Learning》

《Towards Federated Learning at Scale: System Design》

《Differentially Private Learning with Adaptive Clipping》

......(更多联邦学习相关文章请自行搜索 Google Scholar)

Cornell University.

Antonio Marcedone.

《Practical Secure Aggregation for Federated Learning on User-Held Data》

《Practical Secure Aggregation for Privacy-Preserving Machine Learning》

Eugene Bagdasaryan, Andreas Veit, Yiqing Hua, Deborah Estrin, Vitaly Shmatikov.

《How To Backdoor Federated Learning》

《Differential privacy has disparate impact on model accuracy》

Ziteng Sun.

《Can you really backdoor federated learning?》