Federated Averaging Step

Problem

Perform one federated averaging aggregation step.

Return the count-weighted mean of the client updates: sum(counts[k] * updates[k]) / sum(counts). Return [] when the total count is zero or there are no updates.

Hint

Clients are weighted by their number of local samples.

Test cases

case 1

in: [[[1,2],[3,4]],[1,3]]

expected: [2.5,3.5]

case 2

in: [[[1],[3],[5]],[2,1,1]]

expected: [2.5]

case 3

in: [[[2,0]],[5]]

expected: [2,0]

Starter codePython

def federated_average(updates, counts):
    # Your code here
    pass

3 test cases · run your code against them in the editor