Closed roboswell closed 2 months ago
This issue has been marked as stale because it has been open for 365 days with no activity. If this issue is still relevant or if there is new information, please feel free to update or reopen it.
This issue was closed because it has been inactive for 1 days since being marked as stale. If this issue is still relevant or if there is new information, please feel free to update or reopen it.
Fully Homomorphic Encryption (FHE) is extremely slow and thus limits adopting it by the wider Federated Learning community FHE slows down FATE (and all other federated learning frameworks) by a severe rate.
Let's Finally Devise a Way to Speed up FHE. One Solution - The F1 Accelerator The paper, "F1: A Fast and Programmable Accelerator for Fully Homomorphic Encryption" by Axel Feldmann et al. (2021) shows how F1 accelerates FHE and "outperforms state-of-the-art software implementations by gmean 5,400× and by up to 17,000× [and that] these speedups counter most of FHE’s overheads and enable new applications," (pg. 1).
Additional context Though the F1 paper doesn't discuss FHE in the context of Federated Learning, wouldn't it be possible to implement a similar system into the FATE framework to finally begin to resolve the slow speed problem that plagues federated learning?