Filters








4,311 Hits in 6.6 sec

Robotic Grasping using Deep Reinforcement Learning [article]

Shirin Joshi, Sulabh Kumra, Ferat Sahin
2020 arXiv   pre-print
We use the double deep Q-learning framework along with a novel Grasp-Q-Network to output grasp probabilities used to learn grasps that maximize the pick success.  ...  In this work, we present a deep reinforcement learning based method to solve the problem of robotic grasping using visio-motor feedback.  ...  Deep learning has enabled reinforcement learning to be used for decision-making problems such as settings with large dimensional state and action spaces that were once unmanageable.  ... 
arXiv:2007.04499v1 fatcat:l636nzcmqbd2dfek7djyrchlq4

Unsupervised Learning based Jump-Diffusion Process for Object Tracking in Video Surveillance

Xiaobai Liu, Donovan Lo, Chau Thuan
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
from raw videos using reinforcement learning methods.Our method is capable of tracking objects with severe occlusions in crowded scenes and thus recovers the complete trajectories of objects that undergo  ...  This paper presents a principled way for dealing with occlusions in visual tracking which is a long-standing issue in computer vision but largely remains unsolved.  ...  In the discrete action space, for example, one can set an object to be visible if it can be detected with high confidences, or to be occluded if only object parts are detectable (with high confidences)  ... 
doi:10.24963/ijcai.2018/702 dblp:conf/ijcai/LiuLT18 fatcat:hhcrvc2xonhdffbuzxbnmg5jaq

Exploration for Countering the Episodic Memory

Rong Zhou, Yuan Wang, Xiwen Zhang, Chao Wang, Ahmed Mostafa Khalil
2022 Computational Intelligence and Neuroscience  
spaces, or both) since most states occur only once in deep reinforcement learning.  ...  Reinforcement learning is a prominent computational approach for goal-directed learning and decision making, and exploration plays an important role in improving the agent's performance in reinforcement  ...  Deep Q-network (DQN) [2, 3] first attempted to apply reinforcement learning to highdimensional problems by combining Q-learning with deep convolutional neural networks (CNN) as parameterized function  ... 
doi:10.1155/2022/7286186 pmid:35419049 pmcid:PMC8995543 fatcat:3ocvtntowrh2pp6h5uutmpqnfq

Learning Agents with Prioritization and Parameter Noise in Continuous State and Action Space [chapter]

Rajesh Mangannavar, Gopalakrishnan Srinivasaraghavan
2019 Lecture Notes in Computer Science  
We believe these results are a valuable addition to the fast-growing body of results on Reinforcement Learning, more so for continuous state and action space problems.  ...  that can lend themselves naturally to reinforcement based algorithms, and have continuous state and action spaces.  ...  DEEP REINFORCEMENT LEARNING The advent of deep learning has had a significant impact on many areas in machine learning, dramatically improving the state-of-the-art tasks such as object detection, speech  ... 
doi:10.1007/978-3-030-22796-8_22 fatcat:gahibb7yh5fdtglkoow7htapfu

Poisoning Deep Reinforcement Learning Agents with In-Distribution Triggers [article]

Chace Ashcraft, Kiran Karra
2021 arXiv   pre-print
We outline a simple procedure for embedding these, and other, triggers in deep reinforcement learning agents following a multi-task learning paradigm, and demonstrate in three common reinforcement learning  ...  We believe that this work has important implications for the security of deep learning models.  ...  IN-DISTRIBUTION TRIGGER: PURSUIT For a deep multi-agent reinforcement learning (DMARL) example, we consider the Pursuit environment from the PettingZoo library (Terry et al., 2020a), which consists of  ... 
arXiv:2106.07798v1 fatcat:gsplz2ghffc4dbu5xbkyzajqwa

Learning Synergies between Pushing and Grasping with Self-supervised Deep Reinforcement Learning [article]

Andy Zeng, Shuran Song, Stefan Welker, Johnny Lee, Alberto Rodriguez, Thomas Funkhouser
2018 arXiv   pre-print
Skilled robotic manipulation benefits from complex synergies between non-prehensile (e.g. pushing) and prehensile (e.g. grasping) actions: pushing can help rearrange cluttered objects to make space for  ...  In this work, we demonstrate that it is possible to discover and learn these synergies from scratch through model-free deep reinforcement learning.  ...  Our method is based on a pixel-wise version of deep networks that combines deep reinforcement learning with affordance-based manipulation.  ... 
arXiv:1803.09956v3 fatcat:bqotl6d6xvftzfmm7mxe7mx52q

Deep Reinforcement Learning Based Robot Arm Manipulation with Efficient Training Data through Simulation [article]

Xiaowei Xing, Dong Eui Chang
2019 arXiv   pre-print
Deep reinforcement learning trains neural networks using experiences sampled from the replay buffer, which is commonly updated at each time step.  ...  The policy trained with our method works better than the one with the common replay buffer update method. The result is demonstrated both by simulation and by experiment with a real robot arm.  ...  Deep reinforcement learning (DRL), which is the conjunction of deep neural network (DNN) [3] and RL, approximates non-linear multidimensional functions by parameterizing agents' experiences through finite  ... 
arXiv:1907.06884v2 fatcat:2doux7r4rzcq7gv5yt7sl2stdi

Teaching a Robot to Walk Using Reinforcement Learning [article]

Jack Dibachi, Jacob Azoulay
2021 arXiv   pre-print
Deep Q-learning did not yield a high reward policy, often prematurely converging to suboptimal local maxima likely due to the coarsely discretized action space.  ...  Instead, reinforcement learning can train optimal walking policies with ease.  ...  This works well for Markov decision process (MDP) problems with discrete state and action spaces, however is insufficient for problems with large continuous state and action spaces.  ... 
arXiv:2112.07031v1 fatcat:iilpd5wlazf7piwqqxrh4dk6ca

Automated Gain Control Through Deep Reinforcement Learning for Downstream Radar Object Detection [article]

Tristan S.W. Stevens, R. Firat Tigrek, Eric S. Tammam, Ruud J.G. van Sloun
2021 arXiv   pre-print
This paper proposes the use of deep reinforcement learning (RL) to learn policies for gain control under the object detection task.  ...  This approach allows for simulation of vast amounts of data with flexible assignment of the radar parameters to aid in the active learning process.  ...  Deep learning methods for object detection in computer vision can be used for target detection in range-Doppler images [6] , [7] , [8] .  ... 
arXiv:2107.03792v1 fatcat:ebf6l6bwsvhcpl7agz5rood7ke

Shaping the Future through Innovations: From Medical Imaging to Precision Medicine [article]

Dorin Comaniciu, Klaus Engel, Bogdan Georgescu, Tommaso Mansi
2016 arXiv   pre-print
image understanding, support for minimally invasive procedures, and patient-specific computational models with enhanced predictive power.  ...  This manuscript describes recently developed technologies for better handling of image information: photorealistic visualization of medical images with Cinematic Rendering, artificial agents for in-depth  ...  We would also like to thank our clinical and non-clinical collaborators for the strong support over many years. Disclaimer This feature is based on research, and is not commercially available.  ... 
arXiv:1605.02029v2 fatcat:kwsyfx5jqrfuhlblnvnd4pf2n4

Learning Synergies Between Pushing and Grasping with Self-Supervised Deep Reinforcement Learning

Andy Zeng, Shuran Song, Stefan Welker, Johnny Lee, Alberto Rodriguez, Thomas Funkhouser
2018 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
Our method is based on a pixel-wise version of deep networks that combines deep reinforcement learning with affordance-based manipulation.  ...  Naively weighting actions based on visit counts can also be inefficient due our pixel-wise parameterization of the action space.  ... 
doi:10.1109/iros.2018.8593986 dblp:conf/iros/ZengSWLRF18 fatcat:y5jktujbqnf75fmwdr7buhrwum

Vision-Based Robotic Arm Control Algorithm Using Deep Reinforcement Learning for Autonomous Objects Grasping

Hiba Sekkat, Smail Tigani, Rachid Saadane, Abdellah Chehri
2021 Applied Sciences  
This study aims to evolve the grasping task by reaching the intended object based on deep reinforcement learning.  ...  You Only Look Once v5 is employed for object detection, and backward projection is used to detect the three-dimensional position of the target.  ...  Acknowledgments: We would like to thank the anonymous referees for their valuable comments and helpful suggestions.  ... 
doi:10.3390/app11177917 fatcat:ofao3kfp3vhi3bxjhkabvz5xci

Analyzing the Hidden Activations of Deep Policy Networks: Why Representation Matters [article]

Trevor A. McInroe and Michael Spurrier and Jennifer Sieber and Stephen Conneely
2021 arXiv   pre-print
We analyze the hidden activations of neural network policies of deep reinforcement learning (RL) agents and show, empirically, that it's possible to know a priori if a state representation will lend itself  ...  The results from this analysis provide three main insights into how deep RL agents learn.  ...  For a visual example of this training process, see Figure 1 , below. Learning to pick up objects with deep reinforcement learning We deploy all experiments within two distinct tasks.  ... 
arXiv:2103.06398v1 fatcat:ew3bkuxujbbsveme7pfx433d6a

Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation [article]

Tejas D. Kulkarni, Karthik R. Narasimhan, Ardavan Saeedi, Joshua B. Tenenbaum
2016 arXiv   pre-print
Learning goal-directed behavior in environments with sparse feedback is a major challenge for reinforcement learning algorithms.  ...  We present hierarchical-DQN (h-DQN), a framework to integrate hierarchical value functions, operating at different temporal scales, with intrinsically motivated deep reinforcement learning.  ...  We are grateful to receive support from the Center for Brain, Machines and Minds (NSF STC award CCF -1231216) and the MIT OpenMind team.  ... 
arXiv:1604.06057v2 fatcat:p33suojusrcpfpg4ybc4hrfj6y

Bingham Policy Parameterization for 3D Rotations in Reinforcement Learning [article]

Stephen James, Pieter Abbeel
2022 arXiv   pre-print
We propose a new policy parameterization for representing 3D rotations during reinforcement learning.  ...  Today in the continuous control reinforcement learning literature, many stochastic policy parameterizations are Gaussian.  ...  ACKNOWLEDGMENTS This work was supported by the Hong Kong Centre for Logistics Robotics.  ... 
arXiv:2202.03957v1 fatcat:jiftyqwrqzex3k4bd5slsv3bje
« Previous Showing results 1 — 15 out of 4,311 results