Linear Convergence of Forward-Backward Accelerated Algorithms without Knowledge of the Modulus of Strong Convexity