BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:AMLD 2021 Workshop – Federated Learning: collaborative machine l
 earning on sensitive decentralized data
DTSTART:20210929T090000
DTEND:20210929T130000
DTSTAMP:20260511T073031Z
UID:3d1ef10a3f3f5ff5390d65a4a2860f44f6f94c4ae5c9dd8e450580e2
CATEGORIES:Conferences - Seminars
DESCRIPTION:⚠️ A valid COVID certificate must be presented on site t
 o enter the event. ⚠️\n\nDatasets of interest in many application doma
 ins (e.g. healthcare\, finance data\, manufacturing) contain sensitive or
  private information and cannot easily be shared. Additionally\, such dat
 a frequently belongs to multiple distinct parties and combining it in one
  location would expose a lucrative target to hackers. Therefore\, it is d
 esirable to make use of such data without a need to disclose it or store 
 it in a central location. \n\nUnfortunately\, traditional methods to trai
 n predictive models expect data to be fully accessible and centralized on
  a single server. Research work therefore has to rely on small or artific
 ial datasets that can safely be centralized. As a result\, findings frequ
 ently do not generalize well to real-world datapoints and progress is ham
 pered. \n\nFederated Learning (FL) is a recently introduced paradigm that
  addresses this limitation by training models on decentralized datasets w
 ithout requiring centralized data access. This approach allows multiple d
 istinct parties to collaboratively train predictive models without a need
  to directly share sensitive data. Instead of combining datasets\, FL tra
 ins a model in multiple iterations on data subsets stored in different lo
 cations. In every iteration\, every party owning a data subset downloads 
 a copy of the current model weights. An updated model is computed for eac
 h data subset in a local training step. Per-party model updates are then 
 aggregated (a step that can be centralized\, as it does not require data 
 access) resulting in a single overarching FL update step. \n\nAs an intro
 duction to the workshop\, we will introduce the basic concepts underlying
  FL and discuss a few of the key related topics (e.g. Differencial Privac
 y\, \nModel Encryption). Our focus however\, will be on gaining hands-on 
 experience. We will implement a simple Federated Learning system using te
 nsorflow (tensorflow/federated) and pytorch (PySyft). We will give a quic
 k introduction to all needed libraries and tools at the start.\n\n 
LOCATION:STCC 2 52 https://plan.epfl.ch/?room==STCC%202%2052
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR