Escaping Saddle Points for Zeroth-order Nonconvex Optimization using Estimated Gradient Descent

Re-assign community