Program
The workshop will take place on the 8th of December 2023. The schedule of the day is provided below:
| Local Time | Description |
|---|---|
| 09.00 - 09.15 | Opening remarks (Presented by Mario Almeida) |
| 09.15 - 10.15 | Keynote #1 Building RedPajama Ce Zhang (Together and UChicago) |
| 10.15 - 10.45 | Session #1: Training Frameworks (co-ordinated by Stefanos Laskaridis) - Flamingo: A User-Centric System for Fast and Energy-Efficient DNN Training on Smartphones Sanjay Sri Vallabh Singapuram (University of Michigan - Ann Arbor); Chuheng Hu (Johns Hopkins University); Fan Lai, Chensong Zhang (University of Illinois Urbana-Champaign); Mosharaf Chowdhury (University of Michigan - Ann Arbor) - MetisFL: An Embarrassingly Parallelized Controller for Scalable & Efficient Federated Learning Workflows Dimitris Stripelis, Chrysovalantis Anastasiou, Patrick Toral, Armaghan Asghar, Jose Luis Ambite (University of Southern California) |
| 10.45 - 11.15 | Break |
| 11.15 - 12.00 | Session #2: System Heterogeneous Federated Learning (co-ordinated by Alexey Tumanov) - Kimad: Adaptive Gradient Compression with Bandwidth Awareness Jihao Xin, Ivan Ilin (KAUST); Shunkang Zhang (HKUST); Marco Canini, Peter Richtárik (KAUST) - Lightweight Workloads in Heterogeneous Federated Learning via Few-shot Learning Hongrui Shi, Valentin Radu, Po Yuang (University of Sheffield) - Adaptive Decentralized Federated Gossip Learning for Resource-Constrained IoT Devices Lars Wulfert (Fraunhofer Institute for Microelectronic Circuits and Systems); Navidreza Asadi, Wen-Yu Chung (Technical University of Munich); Christian Wiede (Fraunhofer Institute for Microelectronic Circuits and Systems); Anton Grabmaier (University of Duisburg-Essen) |
| 12.00 - 13.00 | Keynote #2 Chakra and ASTRA-sim: An open-source ecosystem for advancing co-design for future distributed AI systems Tushar Krishna (Georgia Tech) |
| 13.00 - 14.00 | Lunch Break |
| 14.00 - 15.00 | Keynote #3 Better Privacy Guarantees for Decentralized Federated Learning Aurélien Bellet (Inria) |
| 15.00 - 15.30 | Session #3: Federated Learning Mechanisms (co-ordinated by Mario Almeida) - Federated Learning is Better with Non-Homomorphic Encryption Konstantin Burlachenko (King Abdullah University of Science and Technology); Abdulmajeed Alrowithi (Saudi Data and AI Authority); Fahad Ali Albalawi (Taif University); Peter Richtarik (KAUST) - Server-Side Stepsizes and Sampling Without Replacement Provably Help in Federated Optimization Grigory Malinovsky (King Abdullah University of Science and Technology); Konstantin Mishchenko (Samsung AI Center Cambridge); Peter Richtarik (King Abdullah University of Science and Technology) |
| 15.30 - 16.00 | Break |
| 16.00 - 17.15 | Panel session: The future of ML compute: Challenges and Opportunities (co-ordinated by Stefanos Laskaridis, Alexey Tumanov) Ada Gavrilovska (Georgia tech), Dan Alistarh (MIT & IST Austria), Ce Zhang (Together and UChicago), Tushar Krishna (Georgia Tech) |
| 17.15 - 17.30 | Concluding remarks |