Filters








32,736 Hits in 4.9 sec

Learning Dynamic Belief Graphs to Generalize on Text-Based Games [article]

Ashutosh Adhikari, Xingdi Yuan, Marc-Alexandre Côté, Mikuláš Zelinka, Marc-Antoine Rondeau, Romain Laroche, Pascal Poupart, Jian Tang, Adam Trischler, William L. Hamilton
2021 arXiv   pre-print
In this work, we investigate how an agent can plan and generalize in text-based games using graph-structured representations learned end-to-end from raw text.  ...  Our work demonstrates that the learned graph-based representations help agents converge to better policies than their text-only counterparts and facilitate effective generalization across game configurations  ...  Building dynamic knowledge graphs from text-based games. arXiv preprint arXiv:1910.09532. 13 whereX is the masked mean of X on the first dimension.  ... 
arXiv:2002.09127v4 fatcat:qahpbnzabzeafdxg4d2fk2mauq

Building Dynamic Knowledge Graphs from Text-based Games [article]

Mikuláš Zelinka, Xingdi Yuan, Marc-Alexandre Côté, Romain Laroche, Adam Trischler
2020 arXiv   pre-print
We are interested in learning how to update Knowledge Graphs (KG) from text.  ...  Furthermore, we introduce a new dataset for KG extraction built upon text-based game transitions (over 300k data points). We conduct experiments and discuss the results.  ...  We are also interested in finding ways of transferring learned graphs from one game to another to improve agents' ability to generalize.  ... 
arXiv:1910.09532v3 fatcat:upvx7fuzgzg2hf4u3nv5bplz44

Enhancing Text-based Reinforcement Learning Agents with Commonsense Knowledge [article]

Keerthiram Murugesan, Mattia Atzeni, Pushkar Shukla, Mrinmaya Sachan, Pavan Kapanipathi, Kartik Talamadupula
2020 arXiv   pre-print
In this paper, we consider the recent trend of evaluating progress on reinforcement learning technology by using text-based environments and games as evaluation environments.  ...  We present one such instantiation of agents that use commonsense knowledge from ConceptNet to show promising performance on two text-based environments.  ...  Our work builds on a recently introduced text-based game TextWorld (Côté et al., 2018) . TextWorld is a sandbox learning environment for training and evaluating RL-based agents on textbased games.  ... 
arXiv:2005.00811v1 fatcat:4c2i7k72gjduxczadwyefdon3m

Text-based RL Agents with Commonsense Knowledge: New Challenges, Environments and Baselines

Keerthiram Murugesan, Mattia Atzeni, Pavan Kapanipathi, Pushkar Shukla, Sadhana Kumaravel, Gerald Tesauro, Kartik Talamadupula, Mrinmaya Sachan, Murray Campbell
2021 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Text-based games have emerged as an important test-bed for Reinforcement Learning (RL) research, requiring RL agents to combine grounded language understanding with sequential decision making.  ...  We design a new text-based gaming environment called TextWorld Commonsense (TWC) for training and evaluating RL agents with a specific kind of commonsense knowledge about objects, their attributes, and  ...  Conclusion We created a novel environment (TWC) to evaluate the performance on RL agents on text-based games requiring commonsense knowledge.  ... 
doi:10.1609/aaai.v35i10.17090 fatcat:e6le5mldrrdz3bnfuetpqiqcku

Text-based RL Agents with Commonsense Knowledge: New Challenges, Environments and Baselines [article]

Keerthiram Murugesan, Mattia Atzeni, Pavan Kapanipathi, Pushkar Shukla, Sadhana Kumaravel, Gerald Tesauro, Kartik Talamadupula, Mrinmaya Sachan, Murray Campbell
2020 arXiv   pre-print
Text-based games have emerged as an important test-bed for Reinforcement Learning (RL) research, requiring RL agents to combine grounded language understanding with sequential decision making.  ...  We design a new text-based gaming environment called TextWorld Commonsense (TWC) for training and evaluating RL agents with a specific kind of commonsense knowledge about objects, their attributes, and  ...  Conclusion We created a novel environment (TWC) to evaluate the performance on RL agents on text-based games requiring commonsense knowledge.  ... 
arXiv:2010.03790v1 fatcat:shnfd4e3ebbazktfmhqrqi5bza

The Design of Disciplinarily-Integrated Games as Multirepresentational Systems

Satyugjit S. Virk, Douglas B Clark, Pratim Sengupta
2017 International Journal of Gaming and Computer-Mediated Simulations  
Disciplinarily-integrated games represent a generalizable genre and template for designing games to support science learning with a focus on bridging across formal and phenomenological representations  ...  The authors leverage the literatures on embodied cognition, adaptive scaffolding, representations in science education, and learning from dynamic visualizations to address the challenges, tradeoffs, and  ...  Research has demonstrated that dynamic linking should be varied based on the level of learners' prior knowledge.  ... 
doi:10.4018/ijgcms.2017070103 fatcat:723vpeby2raq3pbdlxmomptgsy

Learning Knowledge Graph-based World Models of Textual Environments [article]

Prithviraj Ammanabrolu, Mark O. Riedl
2021 arXiv   pre-print
This work focuses on the task of building world models of text-based game environments.  ...  Our world model learns to simultaneously: (1) predict changes in the world caused by an agent's actions when representing the world as a knowledge graph; and (2) generate the set of contextually relevant  ...  Related Work We will focus on three main areas of related work: world modeling and model-based reinforcement learning, and world modeling and knowledge graphs in text games, and general (knowledge) graph  ... 
arXiv:2106.09608v2 fatcat:64j5p4bnxrcvrak4zmssx6gn2y

Toward Automated Quest Generation in Text-Adventure Games [article]

Prithviraj Ammanabrolu, William Broniec, Alex Mueller, Jeremy Paul, Mark O. Riedl
2020 arXiv   pre-print
In this paper, we consider the problem of procedurally generating a quest, defined as a series of actions required to progress towards a goal, in a text-adventure game.  ...  Text-adventure games are typically structured as puzzles or quests wherein the player must execute certain actions in a certain order to succeed.  ...  We then attempt to learn how to generate a quest to complete a recipe-as well as how to create the recipe itself-using a large scale knowledge base of recipes.  ... 
arXiv:1909.06283v4 fatcat:o7poie3e7bcn3fiiks4hi2feci

Bringing Stories Alive: Generating Interactive Fiction Worlds [article]

Prithviraj Ammanabrolu, Wesley Cheung, Dan Tu, William Broniec, Mark O. Riedl
2020 arXiv   pre-print
We perform human participant-based evaluations, testing our neural model's ability to extract and fill-in a knowledge graph and to generate language conditioned on it against rule-based and human-made  ...  In this work, we focus on procedurally generating interactive fiction worlds---text-based worlds that players "see" and "talk to" using natural language.  ...  QAit focuses on helping agents learn procedural knowledge through interaction with a dynamic environment.  ... 
arXiv:2001.10161v1 fatcat:qzok6u4xrzevbjxygyod64wdwm

Simulation-Based Game Learning Environments: Building and Sustaining a Fish Tank

Jason Tan, Gautam Biswas
2007 2007 First IEEE International Workshop on Digital Game and Intelligent Toy Enhanced Learning (DIGITEL'07)  
This paper describes an approach, where an exploration-based simulation environment for conducting and observing science experiments is combined with a game environment to provide a learning environment  ...  The next steps are to extend this environment to a more extended "game" environment that includes a sequence of challenges to further motivate the student and expand their learning abilities.  ...  In the first case the graphs were static and included as part of the text resources, in the second case, they were dynamically generated by the students when they ran a simulation experiment.  ... 
doi:10.1109/digitel.2007.44 dblp:conf/digitel/TanB07 fatcat:wwjpyyydlzc6dmqaurzjsilagi

A Survey of Text Games for Reinforcement Learning Informed by Natural Language

Philip Osborne, Heido Nõmm, André Freitas
2022 Transactions of the Association for Computational Linguistics  
Specifically, this survey: 1) introduces the challenges in Text Game Reinforcement Learning problems, 2) outlines the generation tools for rendering Text Games and the subsequent environments generated  ...  Interactive Fiction Games (or Text Games) are one such problem type that offer a set of safe, partially observable environments where natural language is required as part of the Reinforcement Learning  ...  We find that most Text Games methodologies focus on either an agent's ability to learn efficiently or generalize knowledge.  ... 
doi:10.1162/tacl_a_00495 fatcat:bslztajbmfe2hn3oyt4av6v74a

Interactive Machine Comprehension with Dynamic Knowledge Graphs [article]

Xingdi Yuan
2021 arXiv   pre-print
We describe methods that dynamically build and update these graphs during information gathering, as well as neural models to encode graph representations in RL agents.  ...  An agent must interact with an environment sequentially to gather necessary knowledge in order to answer a question.  ...  Learning dynamic belief graphs to generalize on text-based games. In NeurIPS. CoRR, abs/1612.02605. Daily Mail reading comprehension task.  ... 
arXiv:2109.00077v1 fatcat:gozbgg6ctzgbheghqdukvhlfsa

WordCraft: An Environment for Benchmarking Commonsense Agents [article]

Minqi Jiang, Jelena Luketina, Nantas Nardelli, Pasquale Minervini, Philip H. S. Torr, Shimon Whiteson, Tim Rocktäschel
2020 arXiv   pre-print
To better enable research on agents making use of commonsense knowledge, we propose WordCraft, an RL environment based on Little Alchemy 2.  ...  We evaluate several representation learning methods on this new benchmark and propose a new method for integrating knowledge graphs with an RL agent.  ...  Rule-based game environments such as Go or Chess (Silver et al., 2017; , while very hard to master, are based on simple transition dynamics and largely disconnected from entities and concepts encountered  ... 
arXiv:2007.09185v1 fatcat:vjdlr6xyv5dp3lww3w2pl7sm6y

Galore

Abhishek Parakh, Mahadevan Subramaniam
2022 Journal of the Colloquium for Information System Security Education  
The use of customizable learning objects in multiple different formats such as visual, auditory, text, interactive widgets and newly defined learning objects called gamelets have a potential to tremendously  ...  A parameterized environment, called Galore, that integrates such learning objects into a seamless experience based on student learning styles and preferences for teaching difficult counter and intuitive  ...  ACKNOWLEDGEMENT The authors would like to thank Vidya Bommanapally for help with making figures and graphs.  ... 
doi:10.53735/cisse.v9i1.142 fatcat:fcktdxsqsnhqbh4bjlh5skdty4

Generalization to Novel Objects using Prior Relational Knowledge [article]

Varun Kumar Vijay, Abhinav Ganesh, Hanlin Tang, Arjun Bansal
2019 arXiv   pre-print
These results suggest that our network provides a framework for agents to reason over structured knowledge graphs while still leveraging gradient based learning approaches.  ...  approach is able to apply learned relations to novel objects whereas the baseline algorithms fail.  ...  on text corpora, or learn graph attention models over existing large knowledge graphs [42, 43, 44, 45, 46, 47, 48] .  ... 
arXiv:1906.11315v2 fatcat:zauytd4grbdhri4h4w5movbrqu
« Previous Showing results 1 — 15 out of 32,736 results