Filters








27,666 Hits in 5.5 sec

Dynamical System Inspired Adaptive Time Stepping Controller for Residual Network Families [article]

Yibo Yang, Jianlong Wu, Hongyang Li, Xia Li, Tiancheng Shen, Zhouchen Lin
2019 arXiv   pre-print
The correspondence between residual networks and dynamical systems motivates researchers to unravel the physics of ResNets with well-developed tools in numeral methods of ODE systems.  ...  Inspired by our analyses, we develop an adaptive time stepping controller that is dependent on the parameters of the current step, and aware of previous steps.  ...  Extension to Non-residual Networks Our analyses and the development of our time stepping controller are based on the correspondence between residual networks and discrete dynamical systems.  ... 
arXiv:1911.10305v1 fatcat:7zdfndlxsze4vncegp2btd2t6m

Dynamical System Inspired Adaptive Time Stepping Controller for Residual Network Families

Yibo Yang, Jianlong Wu, Hongyang Li, Xia Li, Tiancheng Shen, Zhouchen Lin
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
The correspondence between residual networks and dynamical systems motivates researchers to unravel the physics of ResNets with well-developed tools in numeral methods of ODE systems.  ...  Inspired by our analyses, we develop an adaptive time stepping controller that is dependent on the parameters of the current step, and aware of previous steps.  ...  Extension to Non-residual Networks Our analyses and the development of our time stepping controller are based on the correspondence between residual networks and discrete dynamical systems.  ... 
doi:10.1609/aaai.v34i04.6141 fatcat:p3sabb4yw5hqvfgppp5qg4r5wm

Energy-Storage-Based Low-Frequency Oscillation Damping Control Using Particle Swarm Optimization and Heuristic Dynamic Programming

Xianchao Sui, Yufei Tang, Haibo He, Jinyu Wen
2014 IEEE Transactions on Power Systems  
In this paper, we investigate the integration of a new goal representation heuristic dynamic programming (GrHDP) algorithm to adaptively control ESD to damp inter-area oscillation.  ...  Various simulation studies with residue-based POD controller design, the proposed PSO optimized controller design, and the GrHDP-based controller design over a four-machine-two-area benchmark power system  ...  dynamic programming (GrHDP), has been introduced in this paper to adaptively control the ESD in a real-time manner.  ... 
doi:10.1109/tpwrs.2014.2305977 fatcat:mgteodmdpndmjeahvuco3pij4m

Adaptive Optics control using Model-Based Reinforcement Learning [article]

Jalo Nousiainen, Chang Rajani, Markus Kasper, Tapio Helin
2021 arXiv   pre-print
Reinforcement Learning (RL) presents a new approach for controlling Adaptive Optics (AO) systems for Astronomy.  ...  We formulate the AO control loop as a model-based RL problem (MBRL) and apply it in numerical simulations to a simple Shack-Hartmann Sensor (SHS) based AO system with 24 resolution elements across the  ...  Acknowledgments We thank Arto Klami, Cedric Heritier, Miska Le Louarn for fruitful discussions and guidance on the project. Disclosures The authors declare no conflicts of interest.  ... 
arXiv:2104.13685v1 fatcat:kubsndvx3bhx5gbbkwv4entxzi

ODE-Inspired Network Design for Single Image Super-Resolution

Xiangyu He, Zitao Mo, Peisong Wang, Yang Liu, Mingyuan Yang, Jian Cheng
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
Not only is it interpretable for super-resolution but it provides a reliable guideline on network designs.  ...  Current studies have achieved impressive performance by manually designing deep residual neural networks but overly relies on practical experience.  ...  First, we revisit the similarity between forward Euler method and residual structure by adopting the view of dynamical system, identifying that we can take advantage of ODEs for SISR network designs.  ... 
doi:10.1109/cvpr.2019.00183 dblp:conf/cvpr/HeMWLY019 fatcat:au75kmyuffdhfdj6i6xh2ioz3q

Lattice Dynamical Wavelet Neural Networks Implemented Using Particle Swarm Optimization for Spatio–Temporal System Identification

Hua-Liang Wei, S.A. Billings, Yifan Zhao, Lingzhong Guo
2009 IEEE Transactions on Neural Networks  
By combining an efficient wavelet representation with a coupled map lattice model, a new family of adaptive wavelet neural networks, called lattice dynamical wavelet neural networks (LDWNN), is introduced  ...  for spatio-temporal system identification.  ...  The dynamics of such a system at a specific spatial location and at the present time instant are dependent on the dynamics of the system at other spatial locations and at previous times.  ... 
doi:10.1109/tnn.2008.2009639 pmid:19129037 fatcat:52miibacrnh5tpmf3fnwvmccri

Convolutional Neural Networks combined with Runge-Kutta Methods [article]

Mai Zhu, Bo Chang, Chong Fu
2019 arXiv   pre-print
Since a pre-activation residual network can be deemed an approximation of a time-dependent dynamical system using the forward Euler method, higher order Runge-Kutta methods (RK methods) can be utilized  ...  A convolutional neural network for image classification can be constructed mathematically since it can be regarded as a multi-period dynamical system.  ...  In other words, s and r control depth of the network while dimensionality reduction controls the width of the network.  ... 
arXiv:1802.08831v6 fatcat:i37jmjf2cbf6ng7r3wkjzoe5q4

A chaotic neural network as motor path generator for mobile robotics

Michele Folgheraiter, Giuseppina Gini
2014 2014 IEEE International Conference on Robotics and Biomimetics (ROBIO 2014)  
The recurrent neural network was implemented in Matlab and a software framework was developed to test the performances of the system with the robot dynamic model.  ...  Preliminary results demonstrate the capability of the neural controller to learn period signals in a short period of time allowing adaptation during the robot operation.  ...  to stimulate the dynamic system.  ... 
doi:10.1109/robio.2014.7090308 dblp:conf/robio/FolgheraiterG14 fatcat:qduu3iy4fzhu7isv53ggv4jkqy

BLACK BOX CLOSED LOOP ROBOT MANIPULATOR SYSTEM IDENTIFICATION

Aziz Said, Ashraf Awad
2007 International Conference on Aerospace Sciences and Aviation Technology  
A remarkable improvement in performance, when using long instead of short format for choosing neural network weights and Bias, is appreciated.  ...  A black box system model is identified from the input-output data. Both linear, OE (Output Error) and non-linear structure (multilayer perceptrons neural network) models are treated and applied.  ...  His fields of interests include switchgear & protection, Optimal, adaptive & preview control, Robotics, and Laser Tracker for machining inspection and control. θ ref1 and θ ref2 are shown in Fig. 3.  ... 
doi:10.21608/asat.2007.24112 fatcat:humtadkqm5agrjm3yfvrawwtsi

Fidelity of Nucleotide Incorporation by the RNA-Dependent RNA Polymerase from Poliovirus [chapter]

C.E. Cameron, I.M. Moustafa, J.J. Arnold
2016 The Enzymes  
All of the relevant conformational states of the enzyme are controlled by a network of interacting residues that permits remote-site residues to control active-site function.  ...  poliovirus (PV), as a model system for understanding the chemical, kinetic, thermodynamic, structural, and dynamical mechanisms employed by this class of polymerases.  ...  Acknowledgments For nearly 20 years, C.E.C. and J.J.A. have worked to elucidate the fundamental principles responsible for the specificity, chemistry, and biology of the viral RdRp.  ... 
doi:10.1016/bs.enz.2016.02.002 pmid:27241934 pmcid:PMC5458529 fatcat:fkb5mywhyjekrexiisotwkww2y

Robust fault detection and isolation scheme using fuzzy wavelet network with a hybrid design algorithm

M. Shahriari-kahkeshi
2017 Scientia Iranica. International Journal of Science and Technology  
adaptive threshold for residual evaluation.  ...  Also, online fault isolation process is performed using fuzzy wavelet network models of the faulty system and analyzing the relation between a bank of residuals.  ...  concepts and development of adaptive thresholds for residual evaluation; (ii) On-line fault isolation step that uses FWN models of the faulty system and analyzes the relation between a bank of residuals  ... 
doi:10.24200/sci.2017.4128 fatcat:bina3kjufzhuzbsrymr6nmizam

An Optimal Control Framework for Joint-channel Parallel MRI Reconstruction without Coil Sensitivities [article]

Wanyu Bian, Yunmei Chen, Xiaojing Ye
2022 arXiv   pre-print
We cast the reconstruction network as a structured discrete-time optimal control system, resulting in an optimal control formulation of parameter training where the parameters of the objective function  ...  for pMRI.  ...  Control inspired learning algorithms introduced a new family of network training models which connect with dynamical systems.  ... 
arXiv:2109.09738v2 fatcat:spnoshmwdvg4fmamu7g2s535pm

Nature Inspired Optimization Techniques, A review for FANETs

2020 Sukkur IBA journal of emerging technologies  
Networking advancements introduces Unmanned Aerial System (UAS) while the device that makes this system applicable is known as flying drones or Unmanned Aerial Vehicle (UAV).  ...  The multi UAV plays an important role in choosing a networking architecture for best performance.  ...  Following necessary steps are required for task operation.  ... 
doi:10.30537/sjet.v3i2.716 fatcat:bcliintbgnd3hkbfkwwwszuchy

Deep Learning Alternative to Explicit Model Predictive Control for Unknown Nonlinear Systems [article]

Jan Drgona, Karol Kis, Aaron Tuor, Draguna Vrabie, Martin Klauco
2021 arXiv   pre-print
In the DPC framework, a neural state-space model is learned from time-series measurements of the system dynamics.  ...  We present differentiable predictive control (DPC) as a deep learning-based alternative to the explicit model predictive control (MPC) for unknown nonlinear systems.  ...  For both the dynamics model and control policy, we select the prediction horizon N = 32 steps, which with sampling time T s = 0.25 seconds, corresponds to the 8 seconds time window.  ... 
arXiv:2011.03699v2 fatcat:3snnkpetargbpbbz2dqt2c5tom

Taking FANET to Next Level

Altaf Hussain, Habib Ullah Khan, Shah Nazir, Ijaz Ullah, Tariq Hussain
2022 Advances in Distributed Computing and Artificial Intelligence Journal  
In other words, this is a pilot free ad-hoc network also known as Unmanned Aerial System (UAS) and the component introduced for such a system is known as UAV.  ...  Moreover, setting up to bolster the range and performance of small UAV in ad hoc network lead to emergent evolution with its high stability, quick deployment, and ease-of-use for the formation of the network  ...  If one protocol is adaptable for the dynamic topology to keep countenance communication, then the other one is responsible for throughput in the network, when the network size and node are increased.  ... 
doi:10.14201/adcaij2021104321337 fatcat:7hrjbxtsprglhk5r7jvjibde6u
« Previous Showing results 1 — 15 out of 27,666 results