Making Non-Stochastic Control (Almost) as Easy as Stochastic

Authors: Max Simchowitz

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Our algorithm is based on a novel variant of online Newton step [19], which adapts to the geometry induced by adversarial disturbances, and our analysis hinges on generic regret bounds for certain structured losses in the OCO-with-memory framework [6]. Theorem 3.1 (informal) When the agent knows the dynamics (1.1) (but does not have foreknowledge of disturbances nor the costs ℓt), DRC-ONS has Control Reg T = O( L2 poly(log T)). Theorem 3.2 (informal) When the dyamics are unknown, DRC-ONS with an initial estimation phase attains Control Reg T = e O( L2 T). All proofs are deferred to our appendix
Researcher Affiliation Academia Max Simchowitz EECS Department UC Berkeley Berkeley, CA 94720 msimchow@berkeley.edu
Pseudocode Yes Algorithm 1: Online Semi-Newton Step Semi-ONS(λ, η, C)
Open Source Code No The paper does not provide an unambiguous statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets No This theoretical paper does not conduct empirical studies with datasets, therefore, no information about publicly available training datasets is provided.
Dataset Splits No This theoretical paper does not conduct empirical studies, therefore, no information about training/validation/test dataset splits is provided.
Hardware Specification No This theoretical paper does not conduct empirical experiments and therefore does not specify any hardware used.
Software Dependencies No This theoretical paper does not specify software dependencies with version numbers.
Experiment Setup No This theoretical paper does not conduct empirical experiments and therefore does not provide details on experimental setup like hyperparameters or training settings.