12,217 Hits in 5.2 sec

Neural Architecture Evolution in Deep Reinforcement Learning for Continuous Control [article]

Jörg K.H. Franke, Gregor Köhler, Noor Awad, Frank Hutter
2020 arXiv   pre-print
Current Deep Reinforcement Learning algorithms still heavily rely on handcrafted neural network architectures.  ...  We propose a novel approach to automatically find strong topologies for continuous control tasks while only adding a minor overhead in terms of interactions in the environment.  ...  Introduction The field of Deep Reinforcement Learning (DRL) bears a lot of potential for meta-learning.  ... 
arXiv:1910.12824v3 fatcat:c6z54jty2ff2bmbxjwctefraru

Designing neural networks through neuroevolution

Kenneth O. Stanley, Jeff Clune, Joel Lehman, Risto Miikkulainen
2019 Nature Machine Intelligence  
, architectures and even the algorithms for learning themselves.  ...  Much of recent machine learning has focused on deep learning, in which neural network weights are trained through variants of stochastic gradient descent.  ...  Current neural network research is largely focused on the fields of 'deep learning' 1,2 and 'deep reinforcement learning' 3, 4 .  ... 
doi:10.1038/s42256-018-0006-z fatcat:gkcu2s7bjvhnxotnhexpwpdyzu

Trends in Neural Architecture Search: Towards the Acceleration of Search [article]

Youngkee Kim, Won Joon Yun, Youn Kyu Lee, Soyi Jung, Joongheon Kim
2021 arXiv   pre-print
In this paper, the main research trends of neural architecture search (NAS) are classified as neuro-evolutionary algorithms, reinforcement learning based algorithms, and one-shot architecture search approaches  ...  In modern deep learning research, finding optimal (or near optimal) neural network models is one of major research directions and it is widely studied in many applications.  ...  In this perspective, Neural Architecture Search (NAS) was emerged in modern deep learning research.  ... 
arXiv:2108.08474v1 fatcat:3m7btivslvcwrei2v2k65swrv4

Autonomous Driving using Deep Reinforcement Learning in Urban Environment

Hashim Shakil Ansari, Goutam R
2019 International Journal of Trend in Scientific Research and Development  
Deep Reinforcement Learning has led us to newer possibilities in solving complex control and navigation related tasks.  ...  The paper presents Deep Reinforcement Learning autonomous navigation and obstacle avoidance of selfdriving cars, applied with Deep Q Network to a simulated car an urban environment.  ...  In the context of 'human-level control', presented a deep reinforcement learning approach which redesigned Q-Learning using a deep neural network.  ... 
doi:10.31142/ijtsrd23442 fatcat:7hfam7cuurerziutdqkkputnnu

Deep Reinforcement Learning for Robotic Manipulation-The state of the art [article]

Smruti Amarjyoti
2017 arXiv   pre-print
This has led to the emergence of a new branch of dynamic robot control system called deep r inforcement learning(DRL).  ...  Several novel and recent approaches have also embedded control policy with efficient perceptual representation using deep learning.  ...  Continuous action space algorithms (CAS) Normalized advantage functions (NAF) Gu et al. proposed a model free approach that used Q-learning to plan in continuous action spaces with deep neural networks  ... 
arXiv:1701.08878v1 fatcat:72obbq4b2rhzngawrrspirk7sq

FPGA Architecture for Deep Learning and its application to Planetary Robotics [article]

Pranay Gankidi, Jekan Thangavelautham
2017 arXiv   pre-print
Q-learning is a form of reinforcement learning and it has been efficient in solving certain class of learning problems.  ...  The architectural implementation for a single neuron Q-learning and a more complex Multilayer Perception (MLP) Q-learning accelerator has been demonstrated.  ...  Reinforcement Learning has been in existence for 30 years, however the most recent innovations of combining learning with deep neural networks [6] is shown to be human competitive.  ... 
arXiv:1701.07543v1 fatcat:ai4uujhxg5exjiipssy65giffi

Hyperparameter Tuning for Deep Reinforcement Learning Applications [article]

Mariam Kiran, Melis Ozyildirim
2022 arXiv   pre-print
Our results are imperative to advance deep reinforcement learning controllers for real-world problems.  ...  In comparison to other neural network architectures, deep RL has not witnessed much hyperparameter tuning, due to its algorithm complexity and simulation platforms needed.  ...  With an optimal neural network architecture, a robust deep RL solution can replace controllers in complex environments for model-free optimum control of various complex systems [6] .  ... 
arXiv:2201.11182v1 fatcat:ilhx5djtlzbcdcohcax6mj5dda

Neural Architecture Search in Embedding Space [article]

Chun-Ting Liu
2020 arXiv   pre-print
The neural architecture search (NAS) algorithm with reinforcement learning can be a powerful and novel framework for the automatic discovering process of neural architectures.  ...  Unlike other NAS with reinforcement learning approaches that search over a discrete and high-dimensional architecture space, this approach enables reinforcement learning to search in an embedding space  ...  Furthermore, manually designing a neural network architecture requires substantial experience in deep learning.  ... 
arXiv:1909.03615v3 fatcat:6kskqjga5fcsbkvdcixjbyvuke

Evolving deep unsupervised convolutional networks for vision-based reinforcement learning

Jan Koutník, Juergen Schmidhuber, Faustino Gomez
2014 Proceedings of the 2014 conference on Genetic and evolutionary computation - GECCO '14  
This is the first use of deep learning in the context evolutionary RL.  ...  Dealing with high-dimensional input spaces, like visual input, is a challenging task for reinforcement learning (RL).  ...  Acknowledgments This research was supported by Swiss National Science Foundation grant #138219: "Theory and Practice of Reinforcement Learning 2", and EU FP7 project: "NAnoSCale Engineering for Novel Computation  ... 
doi:10.1145/2576768.2598358 dblp:conf/gecco/KoutnikSG14 fatcat:5ew6mz3mlnfctnpewtcpirxjfe

Neuroevolution of recurrent architectures on control tasks

Maximilien Le Clei, Pierre Bellec
2022 Proceedings of the Genetic and Evolutionary Computation Conference Companion  
However, constructing network architectures from elementary evolution rules has not yet been shown to scale to modern reinforcement learning benchmarks.  ...  We implement a massively parallel evolutionary algorithm and run experiments on all 19 OpenAI Gym state-based reinforcement learning control tasks.  ...  at optimizing deep neural networks on various reinforcement learning problems [2, 3, 5] .  ... 
doi:10.1145/3520304.3529052 fatcat:a3mq3s7rgvetfmsd7zfqjf74jq

Reinforcement Evolutionary Learning Method for self-learning [article]

Kumarjit Pathak, Jitin Kapila
2018 arXiv   pre-print
Quantitative research is the most widely spread application of data science in Marketing or financial domain where applicability of state of the art reinforcement learning for auto-learning is less explored  ...  Reinforcement learning is heavily dependent on having a simulated environment which is majorly available for gaming or online systems, to learn from the live feedback.  ...  is our primary assumption for using "Deep Policy" method of reinforcement learning.  ... 
arXiv:1810.03198v1 fatcat:xdrufyoocbe5ji34pdbdvcn52y

Systems Modeling Using Deep Elman Neural Network

L. Belhaj Salah, F. Fourati
2019 Engineering, Technology & Applied Science Research  
In this paper, the modeling of complex systems using deep Elman neural network architecture is improved.  ...  Simulation results prove the ability and the efficiency of a deep Elman neural network with two hidden layers in this task.  ...  In [13] , authors have described a combination between deep learning and reinforcement learning for the prediction and control of intelligent laser welding.  ... 
doi:10.48084/etasr.2455 fatcat:azlzbmdh2jge7ct7f2vgbkhj3q

Learning to Evolve [article]

Jan Schuchardt, Vladimir Golkov, Daniel Cremers
2019 arXiv   pre-print
We use deep reinforcement learning to learn to dynamically adjust the strategy of evolutionary algorithms to varying circumstances.  ...  Evolution and learning are two of the fundamental mechanisms by which life adapts in order to survive and to transcend limitations.  ...  ACKNOWLEDGMENTS The authors would like to thank Paolo Notaro for valuable discussions.  ... 
arXiv:1905.03389v1 fatcat:g5nc2ziejnhp3bpgvireujgdle

Online Evolution of Deep Convolutional Network for Vision-Based Reinforcement Learning [chapter]

Jan Koutník, Jürgen Schmidhuber, Faustino Gomez
2014 Lecture Notes in Computer Science  
Dealing with high-dimensional input spaces, like visual input, is a challenging task for reinforcement learning (RL).  ...  Neuroevolution (NE), used for continuous RL problems, has to either reduce the problem dimensionality by (1) compressing the representation of the neural network controllers or (2) employing a pre-processor  ...  Acknowledgments This research was supported by Swiss National Science Foundation grant #138219: "Theory and Practice of Reinforcement Learning 2", and EU FP7 project: "NAnoSCaleEngineering for Novel Computation  ... 
doi:10.1007/978-3-319-08864-8_25 fatcat:i7ltnilzkjgndltzx6vc3bipvq

A New Deep Neural Architecture Search Pipeline for Face Recognition

Ning Zhu, Zekuan Yu, Caixia Kou
2020 IEEE Access  
In this paper, we first propose a new deep neural architecture search pipeline combined with NAS technology and reinforcement learning strategy into face recognition.  ...  INDEX TERMS Neural architecture search, trainable architecture, reinforcement learning, face recognition, large-scale face dataset.  ...  . • Algorithms based on reinforcement learning: NASNet, BlockQNN [26] , ENAS, MnasNet. • Algorithms based on evolution learning: Hierarchical [27] , AmoebaNet [28] . • Continuous differentiable algorithms  ... 
doi:10.1109/access.2020.2994207 fatcat:652qjs23xrckvhplpcu3grod3i
« Previous Showing results 1 — 15 out of 12,217 results