Workshop on Systems Challenges in Reliable and Secure Federated Learning

Co-located with ACM SOSP 2021

October 25, 2021

In Virtual Land

Workshop Schedule

Best paper award goes to:
Fan Lai, Yinwei Dai, Xiangfeng Zhu, Harsha V. Madhyastha, Mosharaf Chowdhury (University of Michigan), "FedScale: Benchmarking Model and System Performance of Federated Learning"  

All peer-reviewed papers are available on ACM DL as Open Access. WWW

Time (US Eastern) Program Item
10:00-10:15 Introductions, best paper award: Organizers [ Video ]

Invited talk: Dinesh Verma, IBM Watson
"Enterprise Challenges in Federated AI Solutions" [ Video ]

10:45-11:15 Invited talk: Reza Shokri, NUS
"ML Privacy Meter" [ Video ]
11:15-11:30 Break

Poster session on Gather of accepted papers (in parallel with other sessions)
Gather event link
Moderator: Kevin Chan (Army Research Lab) 


Peer-reviewed paper session #1 
Session chair: Christof Fetzer (TU Dresden)

[11:30-11:50] Fan Lai (University of Michigan), Yinwei Dai (University of Michigan), Xiangfeng Zhu (University of Michigan), Harsha V. Madhyastha (University of Michigan), Mosharaf Chowdhury (University of Michigan), "FedScale: Benchmarking Model and System Performance of Federated Learning" [ Slides ] [ Video ]
[11:50-12:10] Shuo Liu (Georgetown University), Nirupam Gupta (EPFL), Nitin Vaidya (Georgetown), "Redundancy in cost functions for Byzantine fault-tolerant federated learning" [ Slides ] [ Video ]
[12:10-12:30] Aghiles Ait Messaoud (ESI, Algeria), Vlad Nitu (INSA Lyon), Valerio Schiavoni (University of Neuchatel, Switzerland), Sonia Ben Mokhtar (LIRIS-CNRS, France), "GradSec: a TEE-based Scheme Against Federated Learning Inference Attacks" [ Slides ] [ Video ]

12:30-1:30 Break
1:30-2:00 Invited talk: Do Le Quoc, Huawei Research
"Multi-stakeholder state of Machine Learning" [ Video ]

Panel: FL security: Old wine in new bottle, or NWNB?

Panelists: Salman Avestimehr (USC) [ Video ], Ameet Talwalkar (CMU), Shiva Kasiviswanathan (Amazon) [ Slides ] [ Video ], Gerome Bovet (Armasuisse) [ Slides ] [ Video ]

Moderator: Saurabh Bagchi (Purdue University) 

2:45-3:00 Break

Peer-reviewed paper session #2
Session chair: Suman Jana (Columbia)

[3:00-3:20] Kunlong Liu (University of California, Santa Barbara), Richa Wadaskar (University of California, Santa Barbara), Trinabh Gupta (University of California, Santa Barbara), "Towards an Efficient System for Differentially-private, Cross-device Federated Learning" [ Slides ] [ Video ]
[3:20-3:40] Harikrishna Kuttivelil (University of California, Santa Cruz, CA (UCSC)), Katia Obraczka (University of California, Santa Cruz, CA (UCSC)), "Community-Structured Decentralized Learning for Resilient EI" [ Video ]
[3:40-4:00] Pau-Chen Cheng (IBM Research), Kevin Eykholt (IBM Research), Zhongshu Gu (IBM Research), Hani Jamjoom (IBM Research), K. R. Jayaram (IBM Research), Enriquillo Valdez (IBM Research), Ashish Verma (IBM Research), "Separation of Powers in Federated Learning” [ Slides ] [ Video ]

4:00-4:30 Invited Talk: Andrea Olgiati, AWS
"Debugging Federated Learning"
[ Video ]
4:30-5:00 Invited Talk: Neil Gong, Duke
"Secure Federated Learning" [ Video ]


Workshop Format

We will have three kinds of activities at the workshop:

Invited talks: We will be inviting leading and emerging thinkers in the field to present at the workshop. There will be no publication associated with these talks, but the talks will be recorded and made available through the workshop website.

Poster papers and presentations at the workshop: Each poster paper will be 2 pages in length (plus references). The poster session will be held using Gather Town which gives the feeling of actually moving about in a physical space and multiple attendees interacting concurrently with the authors. This has been used to good effect in forums like NeurIPS 2020.
Gather Event link

Gather instructions:

Panel: We will have a panel focused on a set of closely related topics within the ambit of the workshop. Examples would be:

After the workshop, we will aggregate a selected group of participants and author an article that serves as a review and a vision of the road ahead. This follows in the line of such an article that has been done on the topic from a ML standpoint["Advances and Open Problems in Federated Learning" Kairouz, McMahan et al., December 2019].