Adaptive penalty-based distributed stochastic convex optimization