Filters








214 Hits in 6.7 sec

COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning [article]

Avi Singh, Albert Yu, Jonathan Yang, Jesse Zhang, Aviral Kumar, Sergey Levine
2020 arXiv   pre-print
Reinforcement learning has been applied to a wide variety of robotics problems, but most of such applications involve collecting data from scratch for each new task.  ...  new skills simply through dynamic programming.  ...  This research was supported by the Office of Naval Research, the DARPA Assured Autonomy program, NSF IIS-1651843, and Berkeley DeepDrive, with compute support provided by Amazon, and Google.  ... 
arXiv:2010.14500v1 fatcat:7jql236dcnfknolapuksmvahsq

1GBDT, LR & Deep Learning for Turn-based Strategy Game AI

Like Zhang, Hui Pan, Qi Fan, Changqing Ai, Yanqing Jing
2019 2019 IEEE Conference on Games (CoG)  
In our experiment, we evaluated our model with some other AI strategies (Reinforcement Learning (RL), GBDT, LR, DNN) to against a robot script.  ...  Our research aim is to develop such AI for choosing the logical skills and targets to the player.  ...  We would also like to thank the Tencent K5 Cooperation Department, especially Qi Li, Zuoqiu Shen, Qi wang for comments on the manuscript.  ... 
doi:10.1109/cig.2019.8848103 dblp:conf/cig/ZhangPFAJ19 fatcat:sd3i4gvnmngvfdmbxqhzcbn3pq

DR3: Value-Based Deep Reinforcement Learning Requires Explicit Regularization [article]

Aviral Kumar, Rishabh Agarwal, Tengyu Ma, Aaron Courville, George Tucker, Sergey Levine
2021 arXiv   pre-print
It is reasonable to surmise that deep reinforcement learning (RL) methods could also benefit from this effect.  ...  In this paper, we discuss how the implicit regularization effect of SGD seen in supervised learning could in fact be harmful in the offline deep RL setting, leading to poor generalization and degenerate  ...  Cog: Connecting new skills to past experience with offline reinforcement learning. arXiv preprint arXiv:2010.14500, 2020. Richard S Sutton and Andrew G Barto.  ... 
arXiv:2112.04716v1 fatcat:3jk67c5mpnathc263pzhyy47zi

Electronic 'Knowledge Factories' versus Micro-environment of Innovation: Who Will Win?

Alexandra Moskovskaya
2016 Foresight and STI Governance  
To avoid this, electronic platforms should either learn to recreate the conditions of offline micro-environments of innovation, or not claim to fulfil the role of knowledge production.  ...  experience, but for his/ her communicative capabilities.  ...  Moreover, the connection is stronger when the virtual communication is reinforced with contact in real life [Pan et al., 2015] , which points to the limitations of online communications.  ... 
doi:10.17323/1995-459x.2016.2.81.91 fatcat:spis63zfgjcttcbafn3yt2m5li

Dissolving the Dichotomies Between Online and Campus-Based Teaching: a Collective Response to The Manifesto for Teaching Online (Bayne et al. 2020)

Alison MacKenzie, Alexander Bacalja, Devisakti Annamali, Argyro Panaretou, Prajakta Girme, Maria Cutajar, Sandra Abegglen, Marshall Evens, Fabian Neuhaus, Kylie Wilson, Katerina Psarikidou, Marguerite Koole (+39 others)
2021 Postdigital Science and Education  
As these contributions reveal, the challenges were many and varied, ranging from the positive, breakthrough opportunities that digital learning offered to many students, including the disabled, to the  ...  Though the Manifesto was written before the Covid-19 pandemic, many of the responses gathered here inevitably reflect on the experiences of moving to digital, distant, online teaching under unprecedented  ...  explore new learning paths, encouraged to navigate in their learning quest, and equipped with skills to survive in online learning ecologies.  ... 
doi:10.1007/s42438-021-00259-z fatcat:st2nd4qasracvbgmo7jebtdkia

DEVELOPMENTAL ROBOTICS: THEORY AND EXPERIMENTS

JUYANG WENG
2004 International Journal of Humanoid Robotics  
real-time "living" experiences.  ...  A new Self-Aware Self-Effecting (SASE) agent concept is proposed, based on our SAIL and Dav developmental robots.  ...  Some behavior-based robots, such as Cog § and Kismet¢ at MIT, performed interesting real-time social interactions with humans (some components of Cog were offline learned).  ... 
doi:10.1142/s0219843604000149 fatcat:3bmyboli7jd2lm7ktb2oixaijm

Robust Biped Locomotion Using Deep Reinforcement Learning on Top of an Analytical Control Approach [article]

Mohammadreza Kasaei, Miguel Abreu, Nuno Lau, Artur Pereira, Luis Paulo Reis
2021 arXiv   pre-print
This paper proposes a modular framework to generate robust biped locomotion using a tight coupling between an analytical walking approach and deep reinforcement learning.  ...  This framework is composed of six main modules which are hierarchically connected to reduce the overall complexity and increase its flexibility.  ...  The latter component -a fully connected network -is optimized with reinforcement learning to control the arms residuals and the COM height of the robot, thus improving the upper body efficiency, which  ... 
arXiv:2104.10592v1 fatcat:ulkxmgotivdsrjdnysqimer7b4

Why Heideggerian AI failed and how fixing it would require making it more Heideggerian

Hubert L. Dreyfus
2007 Artificial Intelligence  
To say a hammer has the function of being for hammering leaves out the defining relation of hammers to nails and other equipment, to the point of building things, and to the skills required when actually  ...  Unfortunately, what has always distinguished AI research from a science is its refusal to face up to and learn from its failures.  ...  When an animal learns to respond to a new odor, there is a shift in all other patterns, even if they are not directly involved with the learning.  ... 
doi:10.1016/j.artint.2007.10.012 fatcat:vo2afwdyr5c3vkdckhwyk6kgfu

Why Heideggerian AI Failed and How Fixing It Would Require Making It More Heideggerian [chapter]

Hubert L. Dreyfus
2012 Heidegger and Cognitive Science  
To say a hammer has the function of being for hammering leaves out the defining relation of hammers to nails and other equipment, to the point of building things, and to the skills required when actually  ...  Unfortunately, what has always distinguished AI research from a science is its refusal to face up to and learn from its failures.  ...  When an animal learns to respond to a new odor, there is a shift in all other patterns, even if they are not directly involved with the learning.  ... 
doi:10.1007/978-1-137-00610-3_2 fatcat:vlpkxnwznvgplpqckoi3ghdl7u

Why Heideggerian AI Failed and How Fixing It Would Require Making It More Heideggerian [chapter]

Hubert L. Dreyfus
2008 The Mechanical Mind in History  
To say a hammer has the function of being for hammering leaves out the defining relation of hammers to nails and other equipment, to the point of building things, and to the skills required when actually  ...  Unfortunately, what has always distinguished AI research from a science is its refusal to face up to and learn from its failures.  ...  When an animal learns to respond to a new odor, there is a shift in all other patterns, even if they are not directly involved with the learning.  ... 
doi:10.7551/mitpress/9780262083775.003.0014 fatcat:xkhxro5vp5frpgtvjjy7lpuhl4

Why Heideggerian AI Failed and How Fixing it Would Require Making it More Heideggerian

Hubert L. Dreyfus
2007 Philosophical Psychology  
To say a hammer has the function of being for hammering leaves out the defining relation of hammers to nails and other equipment, to the point of building things, and to the skills required when actually  ...  Unfortunately, what has always distinguished AI research from a science is its refusal to face up to and learn from its failures.  ...  When an animal learns to respond to a new odor, there is a shift in all other patterns, even if they are not directly involved with the learning.  ... 
doi:10.1080/09515080701239510 fatcat:4jmpiz7dr5b3pjfb64wfqnt2ba

A Survey of Artificial Cognitive Systems: Implications for the Autonomous Development of Mental Capabilities in Computational Agents

David Vernon, Giorgio Metta, Giulio Sandini
2007 IEEE Transactions on Evolutionary Computation  
robotic platforms with the aim of implementing bioinspired models of sensorimotor control.  ...  various paradigms of cognition, addressing cognitivist (physical symbol systems) approaches, emergent systems approaches, encompassing connectionist, dynamical, and enactive systems, and also efforts to  ...  ACKNOWLEDGMENT The authors would like to acknowledge the many helpful comments of the two anonymous referees on earlier versions of this paper.  ... 
doi:10.1109/tevc.2006.890274 fatcat:s3ueaficnjbebkrfnend4ahpau

Knowledge management practice system: Theorising from an international meta-standard

Mark Loon
2017 Journal of Business Research  
Link to official URL: http://doi.  ...  of online and offline KM practices.  ...  use technology to primarily codify knowledge or to use technologies to connect people to exchange tacit knowledge.  ... 
doi:10.1016/j.jbusres.2017.11.022 fatcat:3yxjvpo7nnf7phj3xpr3ba4awy

Solving Real World Problems With Alternate Reality Gaming: Student Experiences in the Global Village Playground Capstone Course Design

Mary J Dondlinger, Julie K McLeod
2015 Interdisciplinary Journal of Problem-based Learning  
The Global Village Playground (GVP) was a capstone learning experience designed to address institutional assessment needs while providing an integrated and authentic learning experience for students aimed  ...  Findings suggest strategies for designing technology-based learning environments to support complex problem solving.  ...  You kind of reinforce yourself to create something new. something that you develop, something that you come up with using your own ideas, and use them to create as your own game" (Les).  ... 
doi:10.7771/1541-5015.1488 fatcat:yapvecj3jnczlnf3zsahw55pf4

A Survey of Planning and Learning in Games

Fernando Fradique Duarte, Nuno Lau, Artur Pereira, Luis Paulo Reis
2020 Applied Sciences  
Planning and learning, two well-known and successful paradigms of artificial intelligence, have greatly contributed to these achievements.  ...  This paper presents a survey of the multiple methodologies that have been proposed to integrate planning and learning in the context of games.  ...  Similarly to CBR, the core idea of CBP is to reuse past experience (or cases), in this case encoded in the form of plans, in order to solve new (planning) problems.  ... 
doi:10.3390/app10134529 fatcat:wc27eo2wmvd6lclar7yteyj6cm
« Previous Showing results 1 — 15 out of 214 results