Introduction to Online Nonstochastic Control

• URL: http://arxiv.org/abs/2211.09619v3
• Date: Fri, 19 Jul 2024 00:46:18 GMT
• Title: Introduction to Online Nonstochastic Control
• Authors: Elad Hazan, Karan Singh,
• Abstract summary: In online nonstochastic control, both the cost functions as well as the perturbations from the assumed dynamical model are chosen by an adversary. The target is to attain low regret against the best policy in hindsight from a benchmark class of policies.
• Score: 34.77535508151501
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: This text presents an introduction to an emerging paradigm in control of dynamical systems and differentiable reinforcement learning called online nonstochastic control. The new approach applies techniques from online convex optimization and convex relaxations to obtain new methods with provable guarantees for classical settings in optimal and robust control. The primary distinction between online nonstochastic control and other frameworks is the objective. In optimal control, robust control, and other control methodologies that assume stochastic noise, the goal is to perform comparably to an offline optimal strategy. In online nonstochastic control, both the cost functions as well as the perturbations from the assumed dynamical model are chosen by an adversary. Thus the optimal policy is not defined a priori. Rather, the target is to attain low regret against the best policy in hindsight from a benchmark class of policies. This objective suggests the use of the decision making framework of online convex optimization as an algorithmic methodology. The resulting methods are based on iterative mathematical optimization algorithms, and are accompanied by finite-time regret and computational complexity guarantees.

Related papers

• Provably Mitigating Overoptimization in RLHF: Your SFT Loss is Implicitly an Adversarial Regularizer [52.09480867526656]
  We identify the source of misalignment as a form of distributional shift and uncertainty in learning human preferences. To mitigate overoptimization, we first propose a theoretical algorithm that chooses the best policy for an adversarially chosen reward model. Using the equivalence between reward models and the corresponding optimal policy, the algorithm features a simple objective that combines a preference optimization loss and a supervised learning loss.
  arXiv  Detail & Related papers  (2024-05-26T05:38:50Z)
• Optimal Baseline Corrections for Off-Policy Contextual Bandits [61.740094604552475]
  We aim to learn decision policies that optimize an unbiased offline estimate of an online reward metric. We propose a single framework built on their equivalence in learning scenarios. Our framework enables us to characterize the variance-optimal unbiased estimator and provide a closed-form solution for it.
  arXiv  Detail & Related papers  (2024-05-09T12:52:22Z)
• Acceleration in Policy Optimization [50.323182853069184]
  We work towards a unifying paradigm for accelerating policy optimization methods in reinforcement learning (RL) by integrating foresight in the policy improvement step via optimistic and adaptive updates. We define optimism as predictive modelling of the future behavior of a policy, and adaptivity as taking immediate and anticipatory corrective actions to mitigate errors from overshooting predictions or delayed responses to change. We design an optimistic policy gradient algorithm, adaptive via meta-gradient learning, and empirically highlight several design choices pertaining to acceleration, in an illustrative task.
  arXiv  Detail & Related papers  (2023-06-18T15:50:57Z)
• Online Nonstochastic Model-Free Reinforcement Learning [35.377261344335736]
  We investigate robust model robustness guarantees for environments that may be dynamic or adversarial. We provide efficient and efficient algorithms for optimizing these policies. These are the best-known developments in having no dependence on the state-space dimension in having no dependence on the state-space.
  arXiv  Detail & Related papers  (2023-05-27T19:02:55Z)
• A Nonstochastic Control Approach to Optimization [26.744354103012448]
  We show how recent methods from control preconditions can overcome the challenge of convex nonity. We can learn a method that attains a similar result in hindsight from a class of methods.
  arXiv  Detail & Related papers  (2023-01-19T06:08:01Z)
• Lessons from AlphaZero for Optimal, Model Predictive, and Adaptive Control [0.0]
  We show that the principal AlphaZero/TDGammon ideas of approximation in value space and rollout apply very broadly to deterministic and optimal control problems. These ideas can be effectively integrated with other important methodologies such as model control, adaptive control, decentralized control, and neural network-based value and policy approximations.
  arXiv  Detail & Related papers  (2021-08-20T19:17:35Z)
• Reinforcement Learning for Adaptive Optimal Stationary Control of Linear Stochastic Systems [15.410124023805249]
  This paper studies the adaptive optimal stationary control of continuous-time linear systems with both additive and multiplicative noises. A novel off-policy reinforcement learning algorithm, named optimistic least-squares-based iteration policy, is proposed.
  arXiv  Detail & Related papers  (2021-07-16T09:27:02Z)
• COMBO: Conservative Offline Model-Based Policy Optimization [120.55713363569845]
  Uncertainty estimation with complex models, such as deep neural networks, can be difficult and unreliable. We develop a new model-based offline RL algorithm, COMBO, that regularizes the value function on out-of-support state-actions. We find that COMBO consistently performs as well or better as compared to prior offline model-free and model-based methods.
  arXiv  Detail & Related papers  (2021-02-16T18:50:32Z)
• Non-stationary Online Learning with Memory and Non-stochastic Control [71.14503310914799]
  We study the problem of Online Convex Optimization (OCO) with memory, which allows loss functions to depend on past decisions. In this paper, we introduce dynamic policy regret as the performance measure to design algorithms robust to non-stationary environments. We propose a novel algorithm for OCO with memory that provably enjoys an optimal dynamic policy regret in terms of time horizon, non-stationarity measure, and memory length.
  arXiv  Detail & Related papers  (2021-02-07T09:45:15Z)
• Logistic Q-Learning [87.00813469969167]
  We propose a new reinforcement learning algorithm derived from a regularized linear-programming formulation of optimal control in MDPs. The main feature of our algorithm is a convex loss function for policy evaluation that serves as a theoretically sound alternative to the widely used squared Bellman error.
  arXiv  Detail & Related papers  (2020-10-21T17:14:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.