Multiobjective Reinforcement Learning: A Comprehensive Overview

By Chunming Liu, Xin Xu, Dewen Hu. 2015

412 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Electrical and Electronic Engineering

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 Computational Theory and Mathematics

In Electrical and Electronic Engineering

Citations in the corpus

Reinforcement Learning in Healthcare: A Survey (2019, background): “Moreover, considering that the medical decisions always involve two or more related or contradictory aspects during treatments such as benefits versus associated cost, efficacy versus toxicity, and efficiency versus risk, it is natural to formulate the learning problem into an MORL problem [135].”
Deep Reinforcement Learning for Intelligent Transportation Systems: A Survey (2020, background): “Multi-objectivity is gaining popularity in RL [95] due to its capabilities in complex environments.”
Urban Driving with Multi-Objective Deep Reinforcement Learning (2018, background): “ where the basis functions are learned online. An overview of multi-objective optmization can be found in [12], and a summary of existing multi-objective reinforcement learning approaches is given in [11]. 3 BACKGROUND 3.1 Multi-Objective Reinforcement Learning Multi-objective reinforcement learning”
A Generalized Algorithm for Multi-Objective Reinforcement Learning and Policy Adaptation (2019, background): “1 Related Work Multi-Objective RL Existing MORL algorithms can be roughly divided into two main categories [14, 15, 2]: single-policy methods and multiple-policy methods.”
A Distributional View on Multi-Objective Policy Optimization (2020, background): “Multi-objective reinforcement learning (MORL) algorithms aim to tackle such problems (Roijers et al., 2013; Liu et al., 2015).”
Scaling up budgeted reinforcement learning (2019, background): “rcement Learning also assumes that the performance can be described solely by a reward function. Conversely, real problems typically involve many aspects, some of which can be contradictory [Chunming Liu et al., 2014]. For instance, a self-driving car needs to balance between progressing quickly on ”
Multi-objective optimization of radiotherapy: distributed Q-learning and agent-based simulation (2017, background): “Reinforcement learning (RL) is a major class of machine learning techniques (Liu, Xu, & Hu, 2015).”
A Survey of Preference-Based Reinforcement Learning Methods (2017, background): “A set of Pareto-optimal policies could then be learned using multiobjective reinforcement learning (Liu et al., 2015).”
Toward Edge Intelligence: Multiaccess Edge Computing for 5G and Internet of Things (2020, background): “Though the fast-developing AI technologies, such as deep learning [151]–[154] and reinforcement learning [155]–[157], can help solve the complex edge intelligence problem by learning from a large amount of historical data and react efﬁciently to the dynamic scenarios, how to apply AI to make…”
Reinforcement Learning With Task Decomposition for Cooperative Multiagent Systems (2020, background): “Generally, the optimal policies for problem (6) that maximize each subobjective simultaneously does not always exist [32], but one can get a Pareto optimal policy π i ∗ [32]–[35].”