Price of Stability in Quality-Aware Federated Learning
This addresses a key incentive problem in federated learning for clients with noisy data, but it is incremental as it builds on existing label denoising work by adding game-theoretic analysis.
The paper tackles the problem of self-interested clients in federated learning not applying costly label denoising strategies due to heterogeneous valuations, which degrades global model accuracy. It models this as a label denoising game, analyzes the price of stability, and shows that equilibrium outcomes lead to lower accuracy than socially optimal solutions, with numerical experiments on MNIST indicating the price increases with noisier data.
Federated Learning (FL) is a distributed machine learning scheme that enables clients to train a shared global model without exchanging local data. The presence of label noise can severely degrade the FL performance, and some existing studies have focused on algorithm design for label denoising. However, they ignored the important issue that clients may not apply costly label denoising strategies due to them being self-interested and having heterogeneous valuations on the FL performance. To fill this gap, we model the clients' interactions as a novel label denoising game and characterize its equilibrium. We also analyze the price of stability, which quantifies the difference in the system performance (e.g., global model accuracy, social welfare) between the equilibrium outcome and the socially optimal solution. We prove that the equilibrium outcome always leads to a lower global model accuracy than the socially optimal solution does. We further design an efficient algorithm to compute the socially optimal solution. Numerical experiments on MNIST dataset show that the price of stability increases as the clients' data become noisier, calling for an effective incentive mechanism.