Phoenix: A Federated Generative Diffusion Model (Flower Monthly 2023-08)
"Phoenix: A Federated Generative Diffusion Model" by Fiona Victoria (Master's Graduate at the University of Washington Bothell)
LinkedIn: / fionavictoria
Introducing Phoenix, the first unconditional diffusion model (DDPM) using federated learning in Flower. Phoenix leverages strategies to improve the data diversity of generated samples even when trained on data with statistical heterogeneity (Non-IID data).
This talk was part of the Flower Monthly on 3 August 2023. Watch the full recording here: • Flower Monthly 2023-08
Do you want to join the next Flower Monthly? Go to the Flower Monthly website or join the community on Slack to find out when the next Flower Monthly will take place. Have you developed something interesting with Flower and would like to share your project? Contact the Flower Monthly organizers on Slack or via mail!
Flower Monthly: https://flower.dev/conf/flower-monthly/
Flower community on Slack: https://flower.dev/join-slack/
Slack channel: "flower-monthly"
Flower Events Calendar: flower.dev/flower-events-calendar
Mail: [email protected]