Given a number of data owners and a level of non-iid we should distribute the data samples among the data owners.
- Take a look to open source/ state-of-the-art FL data sets.
- integrate these implementation into our code.
- ... implement in house approach.
Given a number of data owners and a level of non-iid we should distribute the data samples among the data owners.