Fast Escape, Slow Convergence: Learning Dynamics of Phase Retrieval under Power-Law Data

Open in new window