15,080 Hits in 5.2 sec

Reinforcement Learning over Knowledge Graphs for Explainable Dialogue Intent Mining

Kai Yang, Xinyu Kong, Yafang Wang, Jie Zhang, Gerard De Melo
2020 IEEE Access  
We rely on policy-guided reinforcement learning to identify paths in a graph to confirm concrete paths of inference that serve as interpretable explanations.  ...  The graph is induced based on the multi-turn dialogue user utterances, the intents, i.e., standard queries of the dialogues, and the sub-intents associated with the dialogues.  ...  The reinforcement learning agent starts from a user utterance from the current multi-turn dialogue and searches the knowledge graph iteratively with the goal of obtaining a precise and interpretable path  ... 
doi:10.1109/access.2020.2991257 fatcat:wtgscficrzdozp25zy2arysxpi

Scaling Up Multiagent Reinforcement Learning for Robotic Systems: Learn an Adaptive Sparse Communication Graph [article]

Chuangchuang Sun, Macheng Shen, Jonathan P. How
2020 arXiv   pre-print
The complexity of multiagent reinforcement learning (MARL) in multiagent systems increases exponentially with respect to the agent number.  ...  Then a sparse communication graph in MARL is learned by graph neural networks based on this new attention mechanism.  ...  Moreover, for scenarios with inherent sparsity, it is shown that the sparsity of the learned communication graph is interpretable.  ... 
arXiv:2003.01040v2 fatcat:ch3zgjpd2vcmbjpfis5nm3gglm

The Emergence of Adversarial Communication in Multi-Agent Reinforcement Learning [article]

Jan Blumenkamp, Amanda Prorok
2020 arXiv   pre-print
Recent work has shown the promise of Graph Neural Networks (GNNs) to learn explicit communication strategies that enable complex multi-agent coordination.  ...  We perform experiments on multi-agent coverage and path planning problems, and employ a post-hoc interpretability technique to visualize the messages that agents communicate to each other.  ...  Non-cooperative multi-agent reinforcement learning.  ... 
arXiv:2008.02616v2 fatcat:y2kklr4qinhtrom3kkttf6hdva

Emergence of linguistic conventions in multi-agent reinforcement learning [article]

Dorota Lipowska, Adam Lipowski
2018 PLoS ONE   pre-print
We examine formation of signaling conventions in a framework of a multi-agent reinforcement learning model.  ...  When the network of interactions between agents is a complete graph or a sufficiently dense random graph, a global consensus is typically reached with the emerging language being a nearly unique object-word  ...  Such models could incorporate agents, which, using the reinforcement learning, would try to establish a language reflecting their multi-object and multi-agent world.  ... 
doi:10.1371/journal.pone.0208095 pmid:30496267 pmcid:PMC6264146 arXiv:1811.07208v1 fatcat:ycxcd3rxarea3o2mu5jeqn5n4y

Self-Organized Polynomial-Time Coordination Graphs [article]

Qianlan Yang, Weijun Dong, Zhizhou Ren, Jianhao Wang, Tonghan Wang, Chongjie Zhang
2022 arXiv   pre-print
Coordination graph is a promising approach to model agent collaboration in multi-agent reinforcement learning.  ...  In experiments, we show that our approach learns succinct and well-adapted graph topologies, induces effective coordination, and improves performance across a variety of cooperative multi-agent tasks.  ...  They are combined with multi-agent deep reinforcement learning by recent work (Castellini et al., 2019; Böhmer et al., 2020; Li et al., 2020; Wang et al., 2021b) .  ... 
arXiv:2112.03547v3 fatcat:ebtir4pzabfw3eras6gqdlzkpe

VWA: ViewpointS Web Application to Assess Collective Knowledge Building [chapter]

Philippe Lemoisson, Clarel M. H. Rakotondrahaja, Aroniaina Safidy Précieux Andriamialison, Harish A. Sankar, Stefano A. Cerri
2019 Lecture Notes in Computer Science  
User agents feed the graph with resources and viewpoints and exploit maps where resources are linked by "synapses" aggregating the viewpoints.  ...  We expose the mechanism underlying the reinforcement along the knowledge paths and introduce a new measure called Multi Paths Proximity inspired from the parallel neural circuits in the brain.  ...  Zone 1 lists the menu commands: • New resource: Knowledge graphs (KGs) are populated with different classes of resources: "artificial agent", "human agent", "numeric document", "physical document", or  ... 
doi:10.1007/978-3-030-28377-3_1 fatcat:l6nzs5n6n5enferiw7k2d3b7sy

Deep Implicit Coordination Graphs for Multi-agent Reinforcement Learning [article]

Sheng Li, Jayesh K. Gupta, Peter Morales, Ross Allen, Mykel J. Kochenderfer
2021 arXiv   pre-print
Multi-agent reinforcement learning (MARL) requires coordination to efficiently solve certain tasks.  ...  DICG allows learning the tradeoff between full centralization and decentralization via standard actor-critic methods to significantly improve coordination for domains with large number of agents.  ...  ACKNOWLEDGMENTS A.2 StarCraft II Multi-agent Challenge (SMAC) We use the open source SMAC implementation by Samvelyan et al. [32] . We use the SMAC implementation 4 by Samvelyan et al. [32] .  ... 
arXiv:2006.11438v2 fatcat:7kc4f4xabzc5tg4655prfo326y

Graph Signal Sampling via Reinforcement Learning [article]

Oleksii Abramenko, Alexander Jung
2018 arXiv   pre-print
We formulate the problem of sampling and recovering clustered graph signal as a multi-armed bandit (MAB) problem.  ...  This formulation lends naturally to learning sampling strategies using the well-known gradient MAB algorithm.  ...  We propose a novel approach to the graph signal sampling and recovery it by interpreting it as a reinforcement learning (RL) problem.  ... 
arXiv:1805.05827v1 fatcat:e5z3ldtjy5ambdq24xjgd6rr3m

A Reinforcement Learning Approach for Attack Graph Analysis

Mehdi Yousefi, Nhamo Mtetwa, Yan Zhang, Huaglory Tianfield
2018 2018 17th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/ 12th IEEE International Conference On Big Data Science And Engineering (TrustCom/BigDataSE)  
First, we employ multi-host multi-stage vulnerability analysis (MulVAL) to generate an attack graph for a given network topology.  ...  This paper presents an approximate analysis approach for attack graphs based on Q-learning.  ...  This graph draws all the possible attacker's movements between vulnerabilities in the network and it is simplified to be understood and interpreted easily.  ... 
doi:10.1109/trustcom/bigdatase.2018.00041 dblp:conf/trustcom/YousefiMZT18 fatcat:tb3j6t5c3bfb7jrmlujzzo2xdq

Scalable and Transferable Reinforcement Learning for Multi-Agent Mixed Cooperative–Competitive Environments Based on Hierarchical Graph Attention

Yining Chen, Guanghua Song, Zhenhui Ye, Xiaohong Jiang
2022 Entropy  
To deal with these limitations, we propose a deep reinforcement learning (DRL) based multi-agent coordination control method for mixed cooperative–competitive environments.  ...  To improve scalability and transferability when applying in large-scale multi-agent systems, we construct inter-agent communication and use hierarchical graph attention networks (HGAT) to process the local  ...  We describe some background knowledge of multi-agent reinforcement learning and hierarchical graph attention networks in Section 3.  ... 
doi:10.3390/e24040563 pmid:35455226 pmcid:PMC9033143 fatcat:c7asjyqfcbhltayvzzctqivj3e

Learning Policy Representations in Multiagent Systems [article]

Aditya Grover, Maruan Al-Shedivat, Jayesh K. Gupta, Yura Burda, Harrison Edwards
2018 arXiv   pre-print
Our framework casts agent modeling as a representation learning problem.  ...  Consequently, we construct a novel objective inspired by imitation learning and agent identification and design an algorithm for unsupervised learning of representations of agent policies.  ...  NETWORK ARCHITECTURE Agent policies are parameterized as multi-layer perceptrons (MLPs) with 2 hidden layers of 90 units each.  ... 
arXiv:1806.06464v2 fatcat:4uur53q5fvb3pjxb5v732cal2i

Equilibrium Inverse Reinforcement Learning for Ride-hailing Vehicle Network [article]

Takuma Oda
2021 arXiv   pre-print
In this work, we formulate the problem of passenger-vehicle matching in a sparsely connected graph and proposed an algorithm to derive an equilibrium policy in a multi-agent environment.  ...  Furthermore, we developed a method to learn the driver's reward function transferable to an environment with significantly different dynamics from training data.  ...  In this work, we propose SEIRL (spatial equilibrium inverse reinforcement learning), the first approach to multi-agent behavioral modeling and reward learning in equilibrium on a road network for a ride-hailing  ... 
arXiv:2102.06854v1 fatcat:xy6ldr3rhbcrtgeocqhiwq2m5y

Atari-fying the Vehicle Routing Problem with Stochastic Service Requests [article]

Nicholas D. Kullman, Jorge E. Mendoza, Martin Cousineau, Justin C. Goodson
2019 arXiv   pre-print
We present a new general approach to modeling research problems as Atari-like videogames to make them amenable to recent groundbreaking solution methods from the deep reinforcement learning community.  ...  Deep Reinforcement Learning Reinforcement learning (RL), as defined by Sutton et al. refers to the process through which an agent, sequentially interacting with some environment, learns what to do (that  ...  for agents to interpret.  ... 
arXiv:1911.05922v1 fatcat:xbe2dnl2abhdvd5gcqqvf2f6by

Reinforced Causal Explainer for Graph Neural Networks [article]

Xiang Wang, Yingxin Wu, An Zhang, Fuli Feng, Xiangnan He, Tat-Seng Chua
2022 arXiv   pre-print
To address this challenge, we propose a reinforcement learning agent, Reinforced Causal Explainer (RC-Explainer).  ...  Various attribution methods exploit gradient-like or attention scores as the attributions of edges, then select the salient edges with top attribution scores as the explanation.  ...  To efficiently achieve the idea of causal screening, we devise a reinforcement learning agent, Reinforced Causal Explainer (RC-Explainer).  ... 
arXiv:2204.11028v1 fatcat:yweafkvvenaanfg7ctgbecleju

Hierarchical Reinforcement Learning for Multi-agent MOBA Game [article]

Zhijian Zhang, Haozheng Li, Luo Zhang, Tianyin Zheng, Ting Zhang, Xiong Hao, Xiaoxin Chen, Min Chen, Fangxu Xiao, Wei Zhou
2019 arXiv   pre-print
Agent successfully learns to combat and defeat bronze-level built-in AI with 100 experiments show that our method can create a competitive multi-agent for a kind of mobile MOBA game King of Glory in 5v5  ...  The novelty of this work are: (1) proposing a hierarchical framework, where agents execute macro strategies by imitation learning and carry out micromanipulations through reinforcement learning, (2) developing  ...  Multi-agent Reinforcement Learning in Games Multi-agent reinforcement learning(MARL) has certain advantages over single agent learning.  ... 
arXiv:1901.08004v6 fatcat:fiotznnqlfctldreiwepjay6iu
« Previous Showing results 1 — 15 out of 15,080 results