Filters








75,877 Hits in 15.6 sec

Learning the structure of the world: The adaptive nature of state-space and action representations in multi-stage decision-making [article]

Amir Dezfouli, Benrard W Balleine
2017 bioRxiv   pre-print
Here we show that, in the absence of such a priori knowledge, state and action representations adapt to reflect the structure of the world.  ...  State-space and action representations form the building blocks of decision-making processes in the brain; states map external cues to the current situation of the agent whereas actions provide the set  ...  Learning the value of different actions in various states of the environment is essential for decision-making in 296 multi-stage environments.  ... 
doi:10.1101/211664 fatcat:5dhg26tx3ndu3pdbq7otzeleom

Learning the structure of the world: The adaptive nature of state-space and action representations in multi-stage decision-making

Amir Dezfouli, Bernard W. Balleine, Alireza Soltani
2019 PLoS Computational Biology  
State-space and action representations form the building blocks of decision-making processes in the brain; states map external cues to the current situation of the agent whereas actions provide the set  ...  Here we studied how state and action representations adapt to reflect the structure of the world when such a priori knowledge is not available.  ...  In the current study, using a sequential decision-making task in rats, we sought to investigate whether state-space and action representations adapt to the structure of the world.  ... 
doi:10.1371/journal.pcbi.1007334 pmid:31490932 pmcid:PMC6750884 fatcat:gcvioawy7jao5bb32fratqh43q

Structural learning: Embedding discoveries and the dynamics of production

Antonio Andreoni
2014 Structural Change and Economic Dynamics  
The concept of structural learning is introduced to identify the continuous process of structural adjustment triggered and oriented by existing productive structures at each point in time.  ...  The paper then investigates the tension between structure and agency present in structural learning trajectories, and examines the form it takes in different productive organisations.  ...  forms because non-specialised multi-tasks agents cannot make such products.  ... 
doi:10.1016/j.strueco.2013.09.003 fatcat:tdkd7g54tvdshnlotsrakkky34

Toward Compositional Generalization in Object-Oriented World Modeling [article]

Linfeng Zhao, Lingzhi Kong, Robin Walters, Lawson L.S. Wong
2022 arXiv   pre-print
Compositional generalization is a critical ability in learning and decision-making.  ...  We focus on the setting of reinforcement learning in object-oriented environments to study compositional generalization in world modeling.  ...  Introduction In learning and decision-making, the goal is to train models and agents that generalize to new data, tasks, and environments.  ... 
arXiv:2204.13661v1 fatcat:hyvnkmg5bfbctottvjadogysqm

Reward Machines: Exploiting Reward Function Structure in Reinforcement Learning

Rodrigo Toro Icarte, Toryn Q. Klassen, Richard Valenzano, Sheila A. McIlraith
2022 The Journal of Artificial Intelligence Research  
so it can exploit the function's internal structure to learn optimal policies in a more sample efficient manner.  ...  In most RL applications, however, users have to program the reward function and, hence, there is the opportunity to make the reward function visible – to show the reward function's code to the RL agent  ...  Acknowledgments We gratefully acknowledge funding from the Natural Sciences and Engineering Research Council of Canada (NSERC), the Canada CIFAR AI Chairs Program, and Microsoft Research.  ... 
doi:10.1613/jair.1.12440 fatcat:kw7fmnwpgfb2rihbiq2rsmrehm

World model learning and inference

Karl Friston, Rosalyn J. Moran, Yukie Nagai, Tadahiro Taniguchi, Hiroaki Gomi, Josh Tenenbaum
2021 Neural Networks  
of the world-model learning and inference.  ...  and to understand intelligence in the context of how humans interact with their world.  ...  Acknowledgements This article was composed based on the talks and discussions at the International Symposium on Artificial Intelligence and Brain Science 2020 held online, supported by KAKENHI Project  ... 
doi:10.1016/j.neunet.2021.09.011 pmid:34634605 fatcat:ovruawnvivgybmd6keltwmdisy

Exploring applications of deep reinforcement learning for real-world autonomous driving systems [article]

Victor Talpaert, Ibrahim Sobh, B Ravi Kiran, Patrick Mannion, Senthil Yogamani, Ahmad El-Sallab, Patrick Perez
2019 arXiv   pre-print
We first provide an overview of the tasks in autonomous driving systems, reinforcement learning algorithms and applications of DRL to AD systems.  ...  In general, DRL is still at its infancy in terms of usability in real-world applications.  ...  RL algorithms may learn estimates of state values, environment models or policies. In real-world applications, simple tabular representations of estimates are not scalable.  ... 
arXiv:1901.01536v3 fatcat:y3gck5rznjglvim4gem5dvb2ue

Designing Grounded Agents: From RoboCup to the Real-World [chapter]

Christopher Stanton
2009 Lecture Notes in Computer Science  
This paper discusses the nature and role of "grounding" in designing programs for controlling autonomous mobile robots.  ...  That is, currently agents tend to be grounded by their designer's understanding of the world, task, and robot.  ...  The process of embedding an artificial agent in an environment involves identifying regularity and structure within the world that can be used for decision-making.  ... 
doi:10.1007/978-3-642-02921-9_54 fatcat:mva3hal4pnfphn4lwh46uu42l4

Language Learning and Integration of Adult Bhutanese Refugees: An Ethnographic Study [chapter]

Subhash Koirala
2018 Structuring the Thesis  
space influences their investment in natural language learning.  ...  Regardless of the level of women's participation in decision making, both men and women considered men as the legitimate decision makers who have the ability to make the right decisions for the entire  ...  Appendix 1 Interview guide for the Bhutanese students in AMEP classes. Background Appendix 2 Ethics approval.  ... 
doi:10.1007/978-981-13-0511-5_23 fatcat:tqpovrb5szd6lgazp2eyeelm24

ToyArchitecture: Unsupervised Learning of Interpretable Models of the World [article]

Jaroslav Vítků, Petr Dluhoš, Joseph Davidson, Matěj Nikl, Simon Andersson, Přemysl Paška, Jan Šinkora, Petr Hlubuček, Martin Stránský, Martin Hyben, Martin Poliak, Jan Feyereisl (+1 others)
2019 arXiv   pre-print
, learning the influence of one's own actions on the world, model-based reinforcement learning, hierarchical planning and plan execution, and symbolic/sub-symbolic integration in general.  ...  On all levels of the system, the representation of the data can be interpreted in both a symbolic and a sub-symbolic manner.  ...  These systems usually have a fixed structure with adaptive parts and are in some cases able to learn from real-world data.  ... 
arXiv:1903.08772v2 fatcat:wnknrw73pfhnpi6zy35pecriom

Delay Aware Universal Notice Network: Real world multi-robot transfer learning

Samuel Beaussant, Sebastien Lengagne, Benoit Thuilot, Olivier Stasse
2021 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
Remerciements This research was supported supported by the French Research Agency ANR through the AIM project.  ...  ) state and the k last actions taken since, k being the delay in timestep.  ...  Action : a t ∈ R n is the target joints position (n being the number of considered joints). However, the vanilla agent was not making any progress with a full access to the action space.  ... 
doi:10.1109/iros51168.2021.9635917 fatcat:tuac4b7lljah3nyjqa7adzz3tu

A Framework for Human-like Behavior in an immersive virtual world

Fons Kuijk, Sigurd Van Broeck, Claude Dareau, Brian Ravenet, Magalie Ochs, Konstantinos Apostolakis, Petros Daras, David Monaghan, Noel E O'Connor, Julie Wall, Ebroul Izquierdo
2013 2013 18th International Conference on Digital Signal Processing (DSP)  
Rather than just being visualized in a 3D space, the virtual characters (autonomous agents as well as avatars representing users) in the immersive environment facilitate social interaction and multi-party  ...  Represented by realistic virtual characters, this framework allows people to feel immersed in an Internet based virtual world in which they can meet and share experiences in a natural way as they can meet  ...  The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. ICT-2011-7-287723 (REVERIE project).  ... 
doi:10.1109/icdsp.2013.6622826 dblp:conf/icdsp/KuijkBDROADMOWI13 fatcat:aryka2lofvcaznosez2qlydoqe

Being There: Putting Brain, Body and World Together Again

Tim van Gelder, Andy Clark
1998 Philosophical Review  
Such a reconception is prefigured in many dynamical analyses of more representation-hungry Page 170 kinds of problem (such as decision making, and planning 28 ) and is a natural continuation of research  ...  phases of the moon and states of the tide) and "artifactual structure" (learning by exposure to symbols representing the states of moon and tide).  ...  flap followed in quick succession by another in the reverse direction produces a strong, sudden thrust well suited to pouncing on prey or making a fast getaway.  ... 
doi:10.2307/2998391 fatcat:6odpcm5iurgi3kickjadxvghzu

From Complex Regions to Complex Worlds

C. S. Holling
2004 Ecology and Society  
The multi-authored book describing the integrative nature of Panarchy is in part a culmination of fifty years of my own research work, together with that of a fine group of friends and colleagues in the  ...  The front-loop phase is more predictable with higher degrees of certainty. In both the natural and social worlds, it maximizes production and accumulation.  ... 
doi:10.5751/es-00612-090111 fatcat:ovfui4ysnze2rmdtjq47qfojbu

Believable and Effective AI Agents in Virtual Worlds

Iskander Umarov, Maxim Mozgovoy
2012 International Journal of Gaming and Computer-Mediated Simulations  
With games being the way in which many people around the world are learning how to use computers, and one of the primary reasons people spend time on computers, questions regarding how to improve the quality  ...  of the virtual worlds (and the AI agents in them) is an interesting challenge and opportunity for our field.  ...  The decision-making system is represented in the following way. A bot's internal state is encoded with a vector of parameters.  ... 
doi:10.4018/jgcms.2012040103 fatcat:gr37lv6cwrdihajr3skstas7qy
« Previous Showing results 1 — 15 out of 75,877 results