Tag
This paper demonstrates that the regret of Bayesian and multiplicative-weights updates satisfies an exact information-accounting identity, decomposing the learner's excess loss into an uncertainty payment and a reduction in information distance to any comparator. The cumulative payment defines intrinsic time, leading to exact adaptive regret decompositions that unify Hedge, Bayesian model averaging, online convex optimization, and other algorithms.
This paper introduces inner product aware quantization methods that preserve inner products with unseen vectors, developing fast and adaptive algorithms with provable guarantees, achieving 2-10x speedup over prior ASQ methods.
This paper introduces PACE-GGM, a differentially private method for covariance estimation that adaptively selects and measures the most informative entries of the empirical covariance matrix, using Gaussian graphical models for reconstruction. It shows improved estimation error over baselines on real-world data, especially in high-dimensional settings.