Meeting Topic

Data Security and Privacy Protection in Federal Learning Tasks

Meeting background

As The Economist puts it, data is the oil of the 21st century. However, with the official implementation of GDPR(General Data Protection Regulation) in the EU in 2018, data flow among different enterprises and individuals becomes difficult, which seriously hampering the value of data. To address this phenomenon of data silos, Google proposed the concept of federated learning in 2016. This model training method enables all data holders to obtain a perfect model without sharing data. In fact, in the federated learning task, model weights are shared among participants . Although this approach does not directly disclose participants' local data, some attackers have shown that we can still reversely construct their local data based on the model weights uploaded by participants, thereby indirectly creating privacy breaches. So how do we respond to such attacks? Welcome to join us in this SIG meeting to discuss this problem and share.

Meeting details

Meeting Time:
2021-08-31 19:15-21:00 ((UTC+08:00)Beijing)

Meeting Link:
https://welink.zhumu.com/j/184434230

Meeting ID:
0184434230
(We recommend that you download the conference software in advance as prompted by the conference link.)

Tips

1. For better conference experience, you are advised to visit (https://mindspore.cn/federated/) to learn about MindSpore's open-source federated learning framework in advance. After the meeting, we will upload the video to the official MindSpore bilibili account.

2. In addition, we welcome those teachers and students of the AI security field (e.g., adversarial sample, federated learning, differential privacy, secure multi-party computing, model interpretability, deepfake, and speech spoofing) actively sign up, share their achievements, find like-minded partners, and solve problems together!

Agenda

Special guest

We also invited the following professors in this meeting, welcome all professors to give guidance!

Professor Song Tao. He works in the Shanghai Jiao Tong University, Department of Computer Science and Engineering. Professor Song is also a member of Shanghai Key Laboratory of Scalable Computing and Systems. His research interests include cloud computing, distributed computing, machine learning, and systems+AI.

Professor Chen Jinyin. She works at the Institute of Cyberspace Security and School of Information Engineering, Zhejiang University of Technology. Her research interests include artificial intelligence and security, data mining, and complex network analysis.

Professor Jin Xin. He works in Beijing Electronic Science and Technology Institute, and is Head of the Visual Computing and Security Lab (Victory-Lab). His research interests include computational aesthetics, computer vision, and artificial intelligence security.

Professor Li Xiaodong. He works in Beijing Electronic Science and Technology Institute, and his research interests include cloud storage security and blind computing (utility-level homomorphic encryption algorithms).

Welcomoe to join us!