AI RESEARCH
DADA: Dual Averaging with Distance Adaptation
arXiv CS.LG
•
ArXi:2501.10258v2 Announce Type: replace-cross We present a novel universal gradient method for solving convex optimization problems. Our algorithm, Dual Averaging with Distance Adaptation (DADA), is based on the classical scheme of dual averaging and dynamically adjusts its coefficients based on observed gradients and the distance between iterates and the starting point, eliminating the need for problem-specific parameters.