Filters








81 Hits in 5.6 sec

A DNN Framework for Learning Lagrangian Drift With Uncertainty [article]

Joseph Jenkins, Adeline Paiement, Yann Ourmières, Julien Le Sommer, Jacques Verron, Clément Ubelmann, Hervé Glotin
2022 arXiv   pre-print
We remove this constraint by presenting a purely data-driven framework for modelling probabilistic drift in flexible environments.  ...  Reconstructions of Lagrangian drift, for example for objects lost at sea, are often uncertain due to unresolved physical phenomena within the data.  ...  Acknowledgements This work has been supported by Ocean Next, Datlas, and ANRT by means of a PhD CIFRE grant attributed to Joseph Jenkins.  ... 
arXiv:2204.05891v1 fatcat:76bm4tux3rdx7oignhhoder55y

Marine Drifting Trajectory Prediction Based on LSTM-DNN Algorithm

Xianbin Li, Kai Wang, Min Tang, Jiangyi Qin, Peng Wu, Tingting Yang, Haichao Zhang, A.H. Alamoodi
2022 Wireless Communications and Mobile Computing  
In this paper, the long short-term memory with dense neural network (LSTM-DNN) is first introduced to calculate marine drifting trajectory.  ...  By introducing DNN into the algorithm, computational accuracy of drifting trajectory can be significantly improved compared with the conventional LSTM-based prediction model.  ...  By using the model for the trajectories of objects drifting at the ocean surface, Blanken et al. proposed a fuzzy number-based framework for quantifying and propagating uncertainties [31] .  ... 
doi:10.1155/2022/7099494 fatcat:nyjz5ug7uzczrhswxbbkpo3kbm

An Integrated Optimization-Learning Framework for Online Combinatorial Computation Offloading in MEC Networks [article]

Xian Li, Liang Huang, Hui Wang, Suzhi Bi, Ying-Jun Angela Zhang
2021 arXiv   pre-print
To address these challenges, this article overviews the existing methods and introduces a novel framework that efficiently integrates model-based optimization and model-free learning techniques.  ...  On the other hand, the uncertainty of future system states makes it hard for the online decisions to satisfy long-term system constraints.  ...  ., 0.156s for DNN-based LyDROO vs. 8.02s for LyCD with 30 MDs.) [15] .  ... 
arXiv:2104.06619v2 fatcat:o7gqxtac7vgchoqpfvpasfciyq

FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing [article]

Feng Yin, Zhidi Lin, Yue Xu, Qinglei Kong, Deshi Li, Sergios Theodoridis, Shuguang Cui
2020 arXiv   pre-print
In this overview paper, data-driven learning model-based cooperative localization and location data processing are considered, in line with the emerging machine learning and big data methods.  ...  Then, we demonstrate various practical use cases that are summarized from a mixture of standard, newly published, and unpublished works, which cover a broad range of location services, including collaborative  ...  Acknowledgement We would like to thank Wenbiao Guo and Ang Xie from Beijing Jiaotong University and Haole Chen from Wuhan University for their kind help on the manuscript.  ... 
arXiv:2003.03697v2 fatcat:ewbfnkwqdjh77mbi7aut3bhvau

Enhanced data efficiency using deep neural networks and Gaussian processes for aerodynamic design optimization [article]

S. Ashwin Renganathan, Romit Maulik and, Jai Ahuja
2020 arXiv   pre-print
We observe that the latter framework (DNN-BO) improves upon the DNN-only based optimization strategy for the same computational cost.  ...  The optimum shape may then be computed by using a gradient-based optimizer coupled with the trained DNN.  ...  Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public  ... 
arXiv:2008.06731v1 fatcat:5aidgj4axvax5mppd2jhznwtza

FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing

Feng Yin, Zhidi Lin, Qinglei Kong, Yue Xu, Deshi Li, Sergios Theodoridis, Shuguang Cui
2020 IEEE Open Journal of Signal Processing  
In this overview paper, data-driven learning model-based cooperative localization and location data processing are considered, in line with the emerging machine learning and big data methods.  ...  Then, we demonstrate various practical use cases that are summarized from a mixture of standard, newly published, and unpublished works, which cover a broad range of location services, including collaborative  ...  learning model, for instance a DNN or a GP model.  ... 
doi:10.1109/ojsp.2020.3036276 fatcat:dqbg6emly5eddde2dpokekxayy

Information Flow in Deep Neural Networks [article]

Ravid Shwartz-Ziv
2022 arXiv   pre-print
As a result, deep networks are often seen as black boxes with unclear interpretations and reliability.  ...  An analytical framework reveals the underlying structure and optimal representations, and a variational framework using deep neural network optimization validates the results.  ...  He was a remarkable man who gave me so much, and I learned a lot from him. An incredible scholar and a lovely person.  ... 
arXiv:2202.06749v2 fatcat:eo3pcousavg3zp5xza57kejjq4

Deep Learning Theory Review: An Optimal Control and Dynamical Systems Perspective [article]

Guan-Horng Liu, Evangelos A. Theodorou
2019 arXiv   pre-print
Our framework fits nicely with supervised learning and can be extended to other learning problems, such as Bayesian learning, adversarial training, and specific forms of meta learning, without efforts.  ...  Attempts from different disciplines to provide a fundamental understanding of deep learning have advanced rapidly in recent years, yet a unified framework remains relatively limited.  ...  The heuristic of decaying the learning rate with training cycle, i.e. ∼ 1/t, typically works well for DNN training, despite its non-convexity.  ... 
arXiv:1908.10920v2 fatcat:rimioom5ofenvdazcx2lke5gu4

A Physics-informed Deep Learning Approach for Minimum Effort Stochastic Control of Colloidal Self-Assembly [article]

Iman Nodozi, Jared O'Leary, Ali Mesbah, Abhishek Halder
2022 arXiv   pre-print
For specificity, we use a univariate stochastic state model from the literature.  ...  The control objective is formulated in terms of steering the state PDFs from a prescribed initial probability measure towards a prescribed terminal probability measure with minimum control effort.  ...  ACKNOWLEDGMENT We are indebted to Lu Lu for helpful discussions on the DeepXDE toolbox [19] .  ... 
arXiv:2208.09182v2 fatcat:ftfcw7kwurektmcefa7xvjah3y

Search for new physics using effective field theory in 13 TeV pp collision events that contain a top quark pair and a boosted Z or Higgs boson [article]

CMS Collaboration
2022 arXiv   pre-print
A data sample containing top quark pairs (tt̅) produced in association with a Lorentz-boosted Z or Higgs boson is used to search for signs of new physics using effective field theory.  ...  The effects of new physics are probed using a framework in which the standard model is considered to be the low-energy effective field theory of a higher energy scale theory.  ...  with a relative uncertainty of 8 and 20%, respectively.  ... 
arXiv:2208.12837v1 fatcat:kxx5cnceurb6bmtca4dhopvljy

Relieving the Plateau: Active Semi-Supervised Learning for a Better Landscape [article]

Seo Taek Kong, Soomin Jeon, Jaewon Lee, Hongseok Lee, Kyu-Hwan Jung
2021 arXiv   pre-print
Active learning (AL) selects unlabeled instances to be annotated by a human-in-the-loop in hopes of better performance with less labeled data.  ...  Equipped with a few theoretical insights, we propose convergence rate control (CRC), an AL algorithm that selects unlabeled data to improve the problem conditioning upon inclusion to the labeled set, by  ...  Furthermore, there is no standard method in modeling a deep neural network's (DNN) uncertainty, and uncertainty-based AL has its own variants ranging from utilizing Bayesian networks (Kirsch et al., 2019  ... 
arXiv:2104.03525v1 fatcat:cdxrqus7bbeofaxa67ocrcqz6i

Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next [article]

Salvatore Cuomo, Vincenzo Schiano di Cola, Fabio Giampaolo, Gianluigi Rozza, Maziar Raissi, Francesco Piccialli
2022 arXiv   pre-print
This novel methodology has arisen as a multi-task learning framework in which a NN must fit observed data while reducing a PDE residual.  ...  Despite the wide range of applications for which PINNs have been used, by demonstrating their ability to be more feasible in some contexts than classical numerical techniques like Finite Element Method  ...  They extend this model to a Bayesian framework to quantify both epistemic and aleatoric uncertainty.  ... 
arXiv:2201.05624v4 fatcat:rezb3ctw3bamtfrswcwxlc2cvy

2021 Index IEEE Transactions on Automation Science and Engineering Vol. 18

2021 IEEE Transactions on Automation Science and Engineering  
The Author Index contains the primary entry for each item, listed under the first author's name.  ...  ., +, TASE Jan. 2021 106-113 Active Learning for Gaussian Process Considering Uncertainties With Application to Shape Control of Composite Fuselage.  ...  ., +, TASE Oct. 2021 1814-1821 Deep learning Condition-Based Monitoring in Variable Machine Running Conditions Using Low-Level Knowledge Transfer With DNN.  ... 
doi:10.1109/tase.2021.3120615 fatcat:ybfn4kfdvjfipbty7z3mocjjci

DeepParticle: learning invariant measure by a deep neural network minimizing Wasserstein distance on data generated from an interacting particle method [article]

Zhongjian Wang, Jack Xin, Zhiwen Zhang
2022 arXiv   pre-print
functions in closed form nor a finite state space for the samples.  ...  We introduce the so called DeepParticle method to learn and generate invariant measures of stochastic dynamical systems with physical parameters based on data computed from an interacting particle method  ...  Our goal here is to study deep learning in a Lagrangian framework of multi-scale PDE problems, coming naturally from our recent work ( [41] , reviewed in section 4.2 later) on a convergent interacting  ... 
arXiv:2111.01356v3 fatcat:atl7vpps4faapnbc3535vaz4ri

Towards Query-Efficient Black-Box Adversary with Zeroth-Order Natural Gradient Descent [article]

Pu Zhao, Pin-Yu Chen, Siyue Wang, Xue Lin
2020 arXiv   pre-print
Various adversarial attacks are proposed to sabotage the learning performance of DNN models.  ...  Despite the great achievements of the modern deep neural networks (DNNs), the vulnerability/robustness of state-of-the-art DNNs raises security concerns in many application domains requiring high reliability  ...  The black-box setting is more consistent with the scenario of "machine-learning-deployedas-a-service" like Google Cloud Vision API.  ... 
arXiv:2002.07891v1 fatcat:faszu2hcpjbyjaqf7pxzufn4ou
« Previous Showing results 1 — 15 out of 81 results