1: \begin{abstract}

2: This paper investigates how to accelerate the convergence of distributed optimization algorithms on nonconvex problems with zeroth-order information available only.

3: We propose a zeroth-order~(ZO) distributed primal-dual stochastic coordinates algorithm equipped with ``powerball'' method to accelerate.

4: We prove that the proposed algorithm has a convergence rate of $\mathcal{O}(\sqrt{p}/\sqrt{nT})$ for general nonconvex cost functions.

5: %

6: %

7: We consider solving the generation of adversarial examples from black-box DNNs problem to compare with the existing state-of-the-art centralized and distributed ZO algorithms.

8: The numerical results demonstrate the faster convergence rate of the proposed algorithm and match the theoretical analysis.

9: 

10: \end{abstract}

11: