Multi-task Reinforcement Learning in Partially Observable Stochastic Environments

By Xuejun Liao, Lawrence Carin. 2009

108 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Renewable Energy, Sustainability and the Environment, Reinforcement Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Electrical and Electronic Engineering

In Renewable Energy, Sustainability and the Environment

In Reinforcement Learning

Citations in the corpus

An Overview of Multi-task Learning (2018, methodology): “In [70], each task is characterized via a re-gionalized policy and a Dirichlet process is used to cluster tasks.”
A Survey on Multi-Task Learning (2017, background): “In [134], a regionalized policy representation is introduced to characterize the behavior of an agent in each task and the Dirichlet process is placed over regionalized policy representations across multiple tasks to cluster tasks.”
Towards Continual Reinforcement Learning: A Review and Perspectives (2020, background): “Bayesian RL: On the other hand, a more ambitious approach would be to try to learn a belief of the unobserved task state directly from the history of environment interactions (Li et al., 2009; Hernandez-Leal, Zhan, Taylor, Sucar, and de Cote, 2017; Majeed and Hutter, 2018).”
Measuring the Distance Between Finite Markov Decision Processes (2016, methodology): “We consider the maze problem in which many transfer learning methods have been applied [8, 9, 20].”
Online Multi-Task Learning for Policy Gradient Methods (2014, background): “Another example is by Li et al. (2009), who developed a model-free multi-task RL method for partially observable environments.”
Adaptive Information Gathering via Imitation Learning (2017, methodology): “An alternative class of approaches is model-free policy improvement ([30]) which has also been applied to POMDPs ([29, 26]).”
A Bayesian Nonparametic Approach for Estimating Individualized Treatment-Response Curves (2016, methodology): “Others have proposed methods for transfer across tasks (e.g., Li et al. (2009), Ammar et al. (2015)) though often assuming that the responses are modeled at discrete times.”
A Bayesian Nonparametric Approach for Estimating Individualized Treatment-Response Curves (2016, methodology): “ to model the behavior of the population and a x number of other GPs are used to model the group-specic deviations from the population. Others have proposed methods for transfer across tasks (e.g., Li et al. (2009), Ammar et al. (2015)) though often assuming that the responses are modeled at discret”
Distributed Multitask Reinforcement Learning with Quadratic Convergence (2018, background): “When data is scarce, learning task policies jointly through multi-task reinforcement learning (MTRL) rather than independently significantly improves performance [4, 19].”
Stick-Breaking Policy Learning in Dec-POMDPs (2015, methodology): “To address this important yet less addressed problem, a global empirical value function extended from the single-agent case [Li et al., 2009], is constructed based on all the action, observation and reward trajectories, and the product of local policies of all agents.”