Filters








25 Hits in 2.7 sec

BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning [article]

Maxime Chevalier-Boisvert, Dzmitry Bahdanau, Salem Lahlou, Lucas Willems, Chitwan Saharia, Thien Huu Nguyen, Yoshua Bengio
2019 arXiv   pre-print
Here, we introduce the BabyAI research platform to support investigations towards including humans in the loop for grounded language learning.  ...  The levels gradually lead the agent towards acquiring a combinatorially rich synthetic language which is a proper subset of English.  ...  We also thank Rachel Samson, Léonard Boussioux and David Yu-Tung Hui for their help in preparing the final version of the paper.  ... 
arXiv:1810.08272v4 fatcat:gyvuowxf7jhl3kairlwhpm4hh4

Teachable Reinforcement Learning via Advice Distillation [article]

Olivia Watkins, Trevor Darrell, Pieter Abbeel, Jacob Andreas, Abhishek Gupta
2022 arXiv   pre-print
We begin by formalizing a class of human-in-the-loop decision making problems in which multiple forms of teacher-provided advice are available to a learner.  ...  We then describe a simple learning algorithm for these problems that first learns to interpret advice, then learns from advice to complete tasks even in the absence of human supervision.  ...  In Fig 3 left , this improvement process would involve a coach in the loop providing action advice or language sub-goals to the agent during learning to coach it towards successfully accomplishing a task  ... 
arXiv:2203.11197v1 fatcat:ftskchtmfvbvbhtq6rkw3dpyjm

Generalizing Emergent Communication [article]

Thomas A. Unger, Elia Bruni
2020 arXiv   pre-print
We converted the recently developed BabyAI grid world platform to a sender/receiver setup in order to test the hypothesis that established deep reinforcement learning techniques are sufficient to incentivize  ...  the emergence of a grounded discrete communication protocol between generalized agents.  ...  Acknowledgments We thank Tim Baumgärtner, Gautier Dagan, Wilker Fereirra Aziz, Dieuwke Hupkes, Bence Keresztury, Mathijs Mul, Diana Rodríguez Luna and Sainbayar Sukhbaatar for offering their help in producing  ... 
arXiv:2001.01772v3 fatcat:ht22begyzjg27e64cmde5bzudy

A Systematic Survey of Text Worlds as Embodied Natural Language Environments [article]

Peter A Jansen
2021 arXiv   pre-print
This systematic survey outlines recent developments in tooling, environments, and agent modeling for Text Worlds, while examining recent trends in knowledge graphs, common sense reasoning, transfer learning  ...  with rich high-level action spaces while controlling for perceptual input.  ...  ., 2019) , or automated generation systems that may or may not have a human-in-the-loop.  ... 
arXiv:2107.04132v1 fatcat:ebdv2ddhurgtxmdk7ztsk2ejee

Interactive Grounded Language Understanding in a Collaborative Environment: IGLU 2021 [article]

Julia Kiseleva and Ziming Li and Mohammad Aliannejadi and Shrestha Mohanty and Maartje ter Hoeve and Mikhail Burtsev and Alexey Skrynnik and Artem Zholus and Aleksandr Panov and Kavya Srinet and Arthur Szlam and Yuxuan Sun and Marc-Alexandre Côté and Katja Hofmann and Ahmed Awadallah and Linar Abdrazakov and Igor Churin and Putra Manggala and Kata Naszadi and Michiel van der Meer and Taewoon Kim
2022 arXiv   pre-print
The primary goal of the competition is to approach the problem of how to build interactive agents that learn to solve a task while provided with grounded natural language instructions in a collaborative  ...  To facilitate research in this direction, we propose IGLU: Interactive Grounded Language Understanding in a Collaborative Environment.  ...  The BabyAI platform (Chevalier-Boisvert et al., 2019) aims to support investigations towards learning to perform language instructions with a simulated human in the loop. demonstrated that using step-by-step  ... 
arXiv:2205.02388v2 fatcat:bhbdtzhqz5gnzmdoyg67h3ngra

SocialAI: Benchmarking Socio-Cognitive Abilities in Deep Reinforcement Learning Agents [article]

Grgur Kovač, Rémy Portelas, Katja Hofmann, Pierre-Yves Oudeyer
2021 arXiv   pre-print
Building embodied autonomous agents capable of participating in social interactions with humans is one of the main challenges in AI.  ...  As a first step, we propose to expand current research to a broader set of core social skills.  ...  All presented experiments were carried out using both A) the computing facilities MCIA (Mésocentre de Calcul Intensif Aquitain) of the Université de Bordeaux and of the Université de Pau et des Pays de  ... 
arXiv:2107.00956v3 fatcat:6jyi3eivtfctbl2vl66se2jy3q

Grounding Hindsight Instructions in Multi-Goal Reinforcement Learning for Robotics [article]

Frank Röder, Manfred Eppe, Stefan Wermter
2022 arXiv   pre-print
An open problem is the sample-inefficiency that stems from the compositionality of natural language, and from the grounding of language in sensory data and actions.  ...  In addition, we also provide an unexpected result: We show that the learning performance of our agent can be improved by one third if, in a sense, the agent learns to talk to itself in a self-supervised  ...  It provides integrated expert demonstrations, causal puzzles, human-in-the-loop capabilities, and a synthetic subset of the English language, called BabyLanguage.  ... 
arXiv:2204.04308v2 fatcat:3ddj2kamx5anpcfev5hk7nm7xe

Trust in Language Grounding: a new AI challenge for human-robot teams [article]

David M. Bossens, Christine Evers
2022 arXiv   pre-print
This survey provides three contributions relating to the newly emerging field of trust in language grounding, including a) an overview of language grounding research in terms of AI technologies, data sets  ...  The challenge of language grounding is to fully understand natural language by grounding language in real-world referents.  ...  Acknowledgements is work has been supported by the UKRI Trustworthy Autonomous Systems Hub, EP/V00784X/1, and was part of the Trustworthy Human-Robot Teams project.  ... 
arXiv:2209.02066v1 fatcat:kaxjamentbfl5jfvtldlabjose

NeurIPS 2021 Competition IGLU: Interactive Grounded Language Understanding in a Collaborative Environment [article]

Julia Kiseleva, Ziming Li, Mohammad Aliannejadi, Shrestha Mohanty, Maartje ter Hoeve, Mikhail Burtsev, Alexey Skrynnik, Artem Zholus, Aleksandr Panov, Kavya Srinet, Arthur Szlam, Yuxuan Sun (+3 others)
2021 arXiv   pre-print
The primary goal of the competition is to approach the problem of how to build interactive agents that learn to solve a task while provided with grounded natural language instructions in a collaborative  ...  Another important aspect of the challenge is the dedication to perform a human-in-the-loop evaluation as a final evaluation for the agents developed by contestants.  ...  The BabyAI platform [Chevalier-Boisvert et al., 2019] aims to support investigations towards learning to perform language instructions with a simulated human in the loop.  ... 
arXiv:2110.06536v2 fatcat:jyyjlgbmpbdtjf6ctqssl2xpf4

Infinite use of finite means: Zero-Shot Generalization using Compositional Emergent Protocols [article]

Rishi Hazra, Sonu Dixit, Sayambhu Sen
2021 arXiv   pre-print
Human language has been described as a system that makes use of finite means to express an unlimited array of thoughts.  ...  Additionally, we introduce gComm, an environment for investigating grounded language acquisition in 2D-grid environments.  ...  This feature has been added to promote future investigation in multi-agent communication with a human teacher in the loop, thus taking a step towards more complex and realistic human-agent interaction.  ... 
arXiv:2012.05011v3 fatcat:ztflc5jpdnao7etvmfrfhspwwu

IGLU 2022: Interactive Grounded Language Understanding in a Collaborative Environment at NeurIPS 2022 [article]

Julia Kiseleva and Alexey Skrynnik and Artem Zholus and Shrestha Mohanty and Negar Arabzadeh and Marc-Alexandre Côté and Mohammad Aliannejadi and Milagro Teruel and Ziming Li and Mikhail Burtsev and Maartje ter Hoeve and Zoya Volovikova and Aleksandr Panov and Yuxuan Sun and Kavya Srinet and Arthur Szlam and Ahmed Awadallah
2022 arXiv   pre-print
The primary goal of the competition is to approach the problem of how to develop interactive embodied agents that learn to solve a task while provided with grounded natural language instructions in a collaborative  ...  Another critical aspect of the challenge is the dedication to perform a human-in-the-loop evaluation as a final evaluation for the agents developed by contestants.  ...  The BabyAI platform [Chevalier-Boisvert et al., 2019] aims to support investigations towards learning to perform language instructions with a simulated human in the loop. demonstrated that using step-by-step  ... 
arXiv:2205.13771v1 fatcat:5a2hp7wuw5f4bmxx33wjoxisie

Towards Ecologically Valid Research on Language User Interfaces [article]

Harm de Vries, Dzmitry Bahdanau, Christopher Manning
2020 arXiv   pre-print
Language User Interfaces (LUIs) could improve human-machine interaction for a wide variety of tasks, such as playing music, getting insights from databases, or instructing domestic robots.  ...  In contrast to traditional hand-crafted approaches, recent work attempts to build LUIs in a data-driven way using modern deep learning methods.  ...  Ideally, this process measures several aspects of human satisfaction through a human-in-the-loop evaluation with users coming from P T .  ... 
arXiv:2007.14435v1 fatcat:zc3uqtpy3zgrbaimmaos33xdwy

Object Files and Schemata: Factorizing Declarative and Procedural Knowledge in Dynamical Systems [article]

Anirudh Goyal, Alex Lamb, Phanideep Gampa, Philippe Beaudoin, Sergey Levine, Charles Blundell, Yoshua Bengio, Michael Mozer
2020 arXiv   pre-print
The resulting architecture is a drop-in replacement conforming to the same input-output interface as normal recurrent networks (e.g., LSTM, GRU) yet achieves substantially better generalization on environments  ...  Black-box models with a monolithic hidden state often fail to apply procedural knowledge consistently and uniformly, i.e., they lack systematicity.  ...  We are very grateful to Google for giving Google Cloud credits used in this project.  ... 
arXiv:2006.16225v5 fatcat:r4iianl4lfffpm75h7iltwwwee

Factorizing Declarative and Procedural Knowledge in Structured, Dynamical Environments

Anirudh Goyal, Alex Lamb, Phanideep Gampa, Philippe Beaudoin, Charles Blundell, Sergey Levine, Yoshua Bengio, Michael Curtis Mozer
2021 International Conference on Learning Representations  
The resulting architecture is a drop-in replacement conforming to the same input-output interface as normal recurrent networks (e.g., LSTM, GRU) yet achieves substantially better generalization on environments  ...  Black-box models with a monolithic hidden state often fail to apply procedural knowledge consistently and uniformly, i.e., they lack systematicity.  ...  In all cases, the first 10 frames of ground truth are fed in (last 6 shown) and then the system is rolled out for the next 30 time steps.  ... 
dblp:conf/iclr/GoyalLGBBLBM21 fatcat:v73lwjnvr5h53c5efdsrshetbe

Help, Anna! Visual Navigation with Natural Multimodal Assistance via Retrospective Curiosity-Encouraging Imitation Learning [article]

Khanh Nguyen, Hal Daumé III
2019 arXiv   pre-print
to direct the agent towards the goals.  ...  An agent solving tasks in a HANNA environment can leverage simulated human assistants, called ANNA (Automatic Natural Navigation Assistants), which, upon request, provide natural language and visual instructions  ...  Babyai: A platform to study the sample efficiency of grounded language learning. In Proceedings of the International Conference on Learning Representations.  ... 
arXiv:1909.01871v6 fatcat:fup3b6tgqbghdesrx5cpdlqz5y
« Previous Showing results 1 — 15 out of 25 results