Agenda
Day 1 - June 17th
Room: SEA-PKV-2-Center of the Universe
8:00 - 9:00 AM Registration/Breakfast
9:00 - 9:30 AM Daniel Ramage and Brendan McMahan - Welcome & Overview
9:30 - 10:30 AM Lightning Talks (Session Chair: Peter Kairouz)
Adria Gascon, Distributed Summation in the Shuffle Model of Differential Privacy
Borja Balle, Privacy Amplification by Mixing
Aurélien Bellet, Fully Decentralized Joint Learning of Personalized Models and Collaboration Graphs
Gauri Joshi, Adaptive Communication Strategies for Local-Update SGD
Salim El Rouayheb, Single Server PIR: A Song of Computation & Information
Han Yu, Incentive Mechanism Design for Federated Learning
Sanmi Koyejo, Asynchrony and Fault-tolerance in Federated ML; Two Vignettes
Justin Hsu, Data Poisoning against Differentially-Private Learners: Attacks and Defenses
Richard Nock, On Noisy Record Linkage and Learning from Distributed Data
Andres Munoz Medina, Differentially Private Covariance Estimation
Ananda Theertha Suresh / Mehryar Mohri, Agnostic Federated Learning
Jayadev Acharya, Distributed Inference Under Information Constraints: Communication, Privacy, and the Role of Shared Randomness
10:30 -11:00 AM Coffee Break
11:00 - 12:15 PM Session 1: Federated Learning and Analytics (Session Chair: Jakub Konecny)
Françoise Beaufays, Gboard: a Showcase for Federated Learning
Phillip Gibbons, Geo-distributed Learning: The Non-IID Quagmire
Martin Jaggi, Decentralized Training with Compressed Communication
Ayfer Ozgur, A Statistical Estimation Approach to Federated Learning
12:15 - 1:45 PM Lunch & Poster Session
1:45 - 2:15 PM Krzysztof Ostrowski - Using TensorFlow Federated for Research
Nati Srebro - Federated Learning: an Optimization Perspective
2:15 - 3:30 PM Session 2: Privacy, Security, and Fairness I (Session Chair: Keith Bonawitz)
Graham Cormode, Local Differential Privacy: Solution or Distraction?
David Evans, Meaningful Privacy for Federated Learning
Aleksandra Korolova, Hybrid Trust models for Democratizing Differential Privacy
Raef Bassily, Private Stochastic Convex Optimization with Optimal Rate
3:30 - 4:00 PM Coffee Break
4:00 - 5:30 PM Breakout Sessions
B1: Making FL & FA more efficient and effective
B2: Expanding the reach of FL to more devices and more problems
B3: Relaxing the core FL assumptions - applications to new settings and scenarios
B4: Robustness, Attacks, and Defenses
B5. Objectives beyond accuracy and utility: Privacy, Security, and Fairness
5:30 - 6:00 PM University Relations Talk
6:00 - 9:00 PM Argosy Dinner and Cruise
Day 2 - June 18th
Room: SEA-PKV-2-Center of the Universe
8:00 - 8:30 AM Registration/breakfast
8:30 - 9:00 AM Keith Bonawitz - Shaping Information, Building for Trust
9:00 - 10:15 AM Session 3: Privacy, Security, and Fairness II (Session Chair: Kunal Talwar)
Ben Hutchinson, Fairness in ML: Challenges and Opportunities for Federated Learning
Li Xiong, Collaborative Aggregation and Analytics with Differential Privacy
Kareem Amin, Bounding User Contributions: A Bias-Variance Trade-off in Differential Privacy
Naman Agarwal, cpSGD: Communication-efficient and differentially-private distributed SGD
10:15 -10:45 AM Coffee Break
10:45 - 12:00PM Session 4: Robustness, Attacks, and Defenses (Session Chair: Stefano Mazzocchi)
Prateek Mittal, Analyzing Federated Learning through an Adversarial Lens
Emiliano De Cristofaro, Exploiting Unintended Feature Leakage in Collaborative Learning
Zaid Harchaoui, Robust and Secure Aggregation for Federated Learning
Vitaly Shmatikov, Integrity Threats to Federated Learning and How to Mitigate Them
12:00 - 1:00 PM Lunch
1:00 - 1:30 PM Stefano Mazzocchi, Peter Kairouz, and Sean Augenstein - Federated Beyond Learning
1:30 - 3:00 PM Session 5: Relaxing the Core Assumptions of Federated Learning (Session Chair: Sean Augenstein)
Ramesh Raskar, Split Learning: A federated learning approach to thin clients
Farinaz Koushanfar / Tara Javidi, Peer to Peer Federated Bayesian Learning
Mehdi Bennis, Wireless Network Intelligence at the Edge
Dawn Song, Decentralized Federated Learning and its Application for Anomaly Detection
Qiang Yang, Federated Transfer Learning
3:00 - 3:30 PM Coffee Break
3:30 - 4:30 PM Editorial Breakout Sessions
B1: Making FL & FA more efficient and effective
B2: Expanding the reach of FL to more devices and more problems
B3: Relaxing the core FL assumptions - applications to new settings and scenarios
B4: Robustness, Attacks, and Defenses
B5: Objectives beyond accuracy and utility: Privacy, Security, and Fairness
4:30 - 5:00 PM Concluding Note and Q&A