Zobrazeno 1 - 10
of 79
pro vyhledávání: '"Zhang, Shangtong"'
In reinforcement learning, classic on-policy evaluation methods often suffer from high variance and require massive online data to attain the desired accuracy. Previous studies attempt to reduce evaluation variance by searching for or designing prope
Externí odkaz:
http://arxiv.org/abs/2410.05655
Policy evaluation estimates the performance of a policy by (1) collecting data from the environment and (2) processing raw data into a meaningful estimate. Due to the sequential nature of reinforcement learning, any improper data-collecting policy or
Externí odkaz:
http://arxiv.org/abs/2410.02226
Autor:
Blaser, Ethan, Zhang, Shangtong
Tabular average reward Temporal Difference (TD) learning is perhaps the simplest and the most fundamental policy evaluation algorithm in average reward reinforcement learning. After at least 25 years since its discovery, we are finally able to provid
Externí odkaz:
http://arxiv.org/abs/2409.19546
Autor:
Wang, Jiuqi, Zhang, Shangtong
Temporal difference (TD) learning with linear function approximation, abbreviated as linear TD, is a classic and powerful prediction algorithm in reinforcement learning. While it is well understood that linear TD converges almost surely to a unique p
Externí odkaz:
http://arxiv.org/abs/2409.12135
To unbiasedly evaluate multiple target policies, the dominant approach among RL practitioners is to run and evaluate each target policy separately. However, this evaluation method is far from efficient because samples are not shared across policies,
Externí odkaz:
http://arxiv.org/abs/2408.08706
In-context learning refers to the learning ability of a model during inference time without adapting its parameters. The input (i.e., prompt) to the model (e.g., transformers) consists of both a context (i.e., instance-label pairs) and a query instan
Externí odkaz:
http://arxiv.org/abs/2405.13861
Stochastic approximation is a class of algorithms that update a vector iteratively, incrementally, and stochastically, including, e.g., stochastic gradient descent and temporal difference learning. One fundamental challenge in analyzing a stochastic
Externí odkaz:
http://arxiv.org/abs/2401.07844
Autor:
Mathieu, Michaël, Ozair, Sherjil, Srinivasan, Srivatsan, Gulcehre, Caglar, Zhang, Shangtong, Jiang, Ray, Paine, Tom Le, Powell, Richard, Żołna, Konrad, Schrittwieser, Julian, Choi, David, Georgiev, Petko, Toyama, Daniel, Huang, Aja, Ring, Roman, Babuschkin, Igor, Ewalds, Timo, Bordbar, Mahyar, Henderson, Sarah, Colmenarejo, Sergio Gómez, Oord, Aäron van den, Czarnecki, Wojciech Marian, de Freitas, Nando, Vinyals, Oriol
StarCraft II is one of the most challenging simulated reinforcement learning environments; it is partially observable, stochastic, multi-agent, and mastering StarCraft II requires strategic planning over long time horizons with real-time low-level ex
Externí odkaz:
http://arxiv.org/abs/2308.03526
Autor:
Qian, Xiaochi, Zhang, Shangtong
Off-policy learning enables a reinforcement learning (RL) agent to reason counterfactually about policies that are not executed and is one of the most important ideas in RL. It, however, can lead to instability when combined with function approximati
Externí odkaz:
http://arxiv.org/abs/2308.01170
Autor:
Liu, Shuze, Zhang, Shangtong
Most reinforcement learning practitioners evaluate their policies with online Monte Carlo estimators for either hyperparameter tuning or testing different algorithmic design choices, where the policy is repeatedly executed in the environment to get t
Externí odkaz:
http://arxiv.org/abs/2301.13734