Filters








11,103 Hits in 8.4 sec

Accelerating Imitation Learning in Relational Domains via Transfer by Initialization [chapter]

Sriraam Natarajan, Phillip Odom, Saket Joshi, Tushar Khot, Kristian Kersting, Prasad Tadepalli
2014 Lecture Notes in Computer Science  
We consider multi-relational environments such as real-time strategy games and use functional-gradient boosting to capture and transfer the models learned in these environments.  ...  The problem of learning to mimic a human expert/teacher from training trajectories is called imitation learning.  ...  Relational Transfer In this work, we extend the previous work in imitation learning by employing the ideas for inductive transfer [13] .  ... 
doi:10.1007/978-3-662-44923-3_5 fatcat:chx6ajasbrh2nmf7qf2m2qpqdy

Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation [article]

Shani Gamrian, Yoav Goldberg
2019 arXiv   pre-print
The visual mapping from the target to the source domain is performed using unaligned GANs, resulting in a control policy that can be further improved using imitation learning from imperfect demonstrations  ...  We show that by separating the visual transfer task from the control policy we achieve substantially better sample efficiency and transfer behavior, allowing an agent trained on the source task to transfer  ...  Acknowledgements We thank Hal Daum III for the helpful discussions on the Imitation Learning algorithm during the development of the work.  ... 
arXiv:1806.07377v6 fatcat:7a5lxvb57rahxbw36di4nccp5q

Efficient Supervision for Robot Learning Via Imitation, Simulation, and Adaptation

Markus Wulfmeier
2019 Künstliche Intelligenz  
In particular, we target three orthogonal fronts: imitation learning, domain adaptation, and transfer from simulation.  ...  Recent successes in machine learning have led to a shift in the design of autonomous systems, improving performance on existing tasks and rendering new applications possible.  ...  In particular, the three complementary methods presented here build on imitation learning, domain adaptation and transfer learning from simulation in order to reduce the effort connected to introducing  ... 
doi:10.1007/s13218-019-00587-0 fatcat:cfbjm2xhmbf2zbspmb7nq7kocq

Autonomously Reusing Knowledge in Multiagent Reinforcement Learning

Felipe Leno Da Silva, Matthew E. Taylor, Anna Helena Reali Costa
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
Hence, agents must learn how to solve tasks via interactions with the environment.  ...  In this paper, we provide a literature review of methods for knowledge reuse in Multiagent Reinforcement Learning.  ...  Similarly, imitating an agent that has a worse performance will probably hamper the learning process instead of accelerating it.  ... 
doi:10.24963/ijcai.2018/774 dblp:conf/ijcai/SilvaTC18 fatcat:h3pun4shxjaklbocxwcfl2mpme

Learning for a Robot: Deep Reinforcement Learning, Imitation Learning, Transfer Learning

Jiang Hua, Liangcai Zeng, Gongfa Li, Zhaojie Ju
2021 Sensors  
Furthermore, deep reinforcement learning, imitation learning, and transfer learning in robot control are discussed in detail.  ...  The survey reveals that the latest research in deep learning and reinforcement learning has paved the way for highly complex tasks to be performed by robots.  ...  Current algorithms solve the problem of designing the reward function to a certain extent and accelerate the rate of learning by initializing strategies based on teaching data.  ... 
doi:10.3390/s21041278 pmid:33670109 pmcid:PMC7916895 fatcat:ehzsevmddfg5zlyc2wms6yuhui

Reinforcement and Imitation Learning for Diverse Visuomotor Skills

Yuke Zhu, Ziyu Wang, Josh Merel, Andrei Rusu, Tom Erez, Serkan Cabi, Saran Tunyasuvunakool, János Kramár, Raia Hadsell, Nando de Freitas, Nicolas Heess
2018 Robotics: Science and Systems XIV  
In experiments, our reinforcement and imitation agent achieves significantly better performances than agents trained with reinforcement learning or imitation learning alone.  ...  We also illustrate that these policies, trained with large visual and dynamics variations, can achieve preliminary successes in zero-shot sim2real transfer.  ...  Inspired by information hiding strategies used in locomotion domains [13, 28] , our discriminator only takes the object-centric features as input while masking out arm-related information.  ... 
doi:10.15607/rss.2018.xiv.009 dblp:conf/rss/Zhu0MRECTKHFH18 fatcat:u6pt5wi6lvgchhezcsn6qntid4

Reinforcement and Imitation Learning for Diverse Visuomotor Skills [article]

Yuke Zhu, Ziyu Wang, Josh Merel, Andrei Rusu, Tom Erez, Serkan Cabi, Saran Tunyasuvunakool, János Kramár, Raia Hadsell, Nando de Freitas, Nicolas Heess
2018 arXiv   pre-print
In experiments, our reinforcement and imitation agent achieves significantly better performances than agents trained with reinforcement learning or imitation learning alone.  ...  We also illustrate that these policies, trained with large visual and dynamics variations, can achieve preliminary successes in zero-shot sim2real transfer.  ...  Inspired by information hiding strategies used in locomotion domains [13, 28] , our discriminator only takes the object-centric features as input while masking out arm-related information.  ... 
arXiv:1802.09564v2 fatcat:7vwziswy25blbdgmb6gklibl5m

One-Shot Domain-Adaptive Imitation Learning via Progressive Learning [article]

Dandan Zhang, Wen Fan, John Lloyd, Chenguang Yang, Nathan Lepora
2022 arXiv   pre-print
In addition, the generalizability to new domains is improved, as demonstrated here with novel background, target container and granule combinations.  ...  deep imitation learning.  ...  Step 3: Transferring Knowledge via Fine-Tuning. Finally, a new database D o T is constructed by combining D r T and D r T , which is used to fine-tune F C (.) in a transfer learning manner.  ... 
arXiv:2204.11251v1 fatcat:su5ve36iafhirirli636bq5xgm

Pastiche Master: Exemplar-Based High-Resolution Portrait Style Transfer [article]

Shuai Yang, Liming Jiang, Ziwei Liu, Chen Change Loy
2022 arXiv   pre-print
Recent studies on StyleGAN show high performance on artistic portrait generation by transfer learning with limited data.  ...  In this paper, we explore more challenging exemplar-based high-resolution portrait style transfer by introducing a novel DualStyleGAN with flexible control of dual styles of the original face domain and  ...  This study is supported under the RIE2020 Industry Alignment Fund -Industry Collaboration Projects (IAF-ICP) Funding Initiative, as well as cash and in-kind contribution from the industry partner(s).  ... 
arXiv:2203.13248v1 fatcat:pc2wurrfyvb2rmpsec425x3omi

Learning on a Budget via Teacher Imitation [article]

Ercument Ilhan, Jeremy Gow, Diego Perez-Liebana
2021 arXiv   pre-print
In this paper, we extend the idea of advice reusing via teacher imitation to construct a unified approach that addresses both advice collection and advice utilisation problems.  ...  Action advising is a framework that provides a flexible way to transfer such knowledge in the form of actions between teacher-student peers.  ...  ACKNOWLEDGMENT This research utilised Queen Mary's Apocrita HPC facility, supported by QMUL Research-IT. http://doi.org/10.5281/zenodo.438045  ... 
arXiv:2104.08440v3 fatcat:zxvmz7dybneqhfb4q62wwwmj6q

Learning by Watching: Physical Imitation of Manipulation Skills from Human Videos [article]

Haoyu Xiong, Quanzhou Li, Yun-Chun Chen, Homanga Bharadhwaj, Samarth Sinha, Animesh Garg
2021 arXiv   pre-print
In this paper, we present Learning by Watching (LbW), an algorithmic framework for policy learning through imitation from a single video specifying the task.  ...  Learning from visual data opens the potential to accrue a large range of manipulation behaviors by leveraging human demonstrations without specifying each of them mathematically, but rather through natural  ...  By this learned reward model, we achieve imitation from human videos via reinforcement learning.  ... 
arXiv:2101.07241v2 fatcat:nvqthzyfnfhzhhsx42o3yi65wu

Neural Circuit Architectural Priors for Embodied Control [article]

Nikhil X. Bhattasali, Anthony M. Zador, Tatiana A. Engel
2022 arXiv   pre-print
In nature, animals are born with highly structured connectivity in their nervous systems shaped by evolution; this innate circuitry acts synergistically with learning mechanisms to provide inductive biases  ...  However, it is unknown the extent to which ANN architectures inspired by neural circuitry can yield useful biases for other domains.  ...  This work was supported by Schmidt Futures (N.X.B.); the G. Harold and Leila Y.  ... 
arXiv:2201.05242v1 fatcat:ehdp3cxbvrf3xp7kxcymwgjoem

Online State Elimination in Accelerated reinforcement Learning

Safreni Candra Sari, Kuspriyanto Kuspriyanto, Ary Setijadi Prihatmanto, Widyawardana Adiprawita
2014 International Journal on Electrical Engineering and Informatics  
Most successes in accelerating RL incorporated internal knowledge or human intervention into the learning system such as reward shaping, transfer learning, parameter tuning, and even heuristics.  ...  This algorithm accelerates the RL learning performance by distinguishing insignificant states from the significant one and then eliminating them from the state space in early learning episodes.  ...  Another approach in accelerating the RL is by applying transfer learning in RL.  ... 
doi:10.15676/ijeei.2014.6.4.3 fatcat:x7b7lqcl7favbloav44qlafqvi

State Elimination in Accelerated Multiagent Reinforcement Learning

Ary Setijadi Prihatmanto, Widyawardana Adiprawita, Safreni Candra Sari, Kuspriyanto Kuspriyanto
2016 International Journal on Electrical Engineering and Informatics  
This paper presents a novel algorithm of Multiagent Reinforcement Learning called State Elimination in Accelerated Multiagent Reinforcement Learning (SEA-MRL), that successfully produces faster learning  ...  without incorporating internal knowledge or human intervention such as reward shaping, transfer learning, parameter tuning, and even heuristics, into the learning system.  ...  Another approach in accelerating the RL is by applying transfer learning in RL.  ... 
doi:10.15676/ijeei.2016.8.3.12 fatcat:aydhzofht5hfhlypek4xkterpq

Trajectory learning from human demonstrations via manifold mapping

Michihisa Hiratsuka, Ndivhuwo Makondo, Benjamin Rosman, Osamu Hasegawa
2016 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
Experiments in simulation on a 4 degree of freedom robot show that our method is able to correctly imitate various skills demonstrated by a human.  ...  The kinematics mapping between the robot and the human model is learned by employing Local Procrustes Analysis, which enables the transfer of the demonstrated trajectory from the human model to the robot  ...  the target domain in order to improve learning of kinematic models in the target domain.  ... 
doi:10.1109/iros.2016.7759579 dblp:conf/iros/HiratsukaMRH16 fatcat:h2kfkkbma5c4dc5laxjlv6k5h4
« Previous Showing results 1 — 15 out of 11,103 results