Question 1

How is total training time calculated?

Accepted Answer

Total training time = (computation time per round + communication time per round) × total target rounds. Computation per round = local epochs × dataset size / 1000 (simulated throughput). Communication per round = model size / bandwidth × 8 (converted to seconds).

Question 2

What strategies are compared?

Accepted Answer

FedAvg (baseline), FedProx (proximal term, ~10% slower convergence), FedMA (matching & averaging, ~15% faster convergence). Estimates are based on round efficiency.

Question 3

What is communication vs computation ratio?

Accepted Answer

Ratio = communication time per round / computation time per round. Higher ratio indicates communication bottleneck.

Strategy	Rounds to converge	Total time (est.)	Speed vs FedAvg
FedAvg	—	—	baseline
FedProx	—	—	—
FedMA	—	—	—

⚡ Federated Learning Round Estimator

📊 Training Overview

🧠 Strategy comparison · Convergence rounds to target accuracy

⏱️ Federated rounds timeline (first 40 rounds · FedAvg · FedProx · FedMA)