Filters








111,367 Hits in 1.9 sec

Learning Similarity Metrics for Numerical Simulations [article]

Georg Kohl, Kiwon Um, Nils Thuerey
2020 arXiv   pre-print
We propose a neural network-based approach that computes a stable and generalizing metric (LSiM) to compare data from a variety of numerical simulation sources.  ...  To demonstrate that the proposed approach outperforms existing metrics for vector spaces and other learned, image-based metrics, we evaluate the different methods on a large range of test data.  ...  We would like to thank Stephan Rasp for preparing the WeatherBench data and all reviewers for helping to improve this work.  ... 
arXiv:2002.07863v2 fatcat:chi2mlzmqjc2bjczbw3e7nyfae

A Pointwise Evaluation Metric to Visualize Errors in Machine Learning Surrogate Models [chapter]

Seyed Shayan Sajjadinia, Bruno Carpentieri, Gerhard A. Holzapfel
2021 Frontiers in Artificial Intelligence and Applications  
Numerical simulation is widely used to study physical systems, although it can be computationally too expensive.  ...  learning progress can be observed.  ...  This metric may be applicable when the points are attached to the simulated materials in the Euclidean space that should have similar locations at some stage, regularly at the beginning of the numerical  ... 
doi:10.3233/faia210386 fatcat:tajvmcw7gjgwdjofc6boxfr4ea

Distance-learning For Approximate Bayesian Computation To Model a Volcanic Eruption

Lorenzo Pacchiardi, Pierre Künzli, Marcel Schöngens, Bastien Chopard, Ritabrata Dutta
2020 Sankhya B  
However, informative summary statistics cannot be found for the considered model. We therefore develop a technique to learn a distance between model outputs based on deep metric-learning.  ...  Using ABC, which depends on many simulations from the considered model, we develop an inferential framework to learn parameters of a stochastic numerical simulator of volcanic eruption.  ...  We thank CADMOS for providing computing resources at the Swiss Super Computing Center.  ... 
doi:10.1007/s13571-019-00208-8 fatcat:46ekiyg2prcoha2upei2mcdeke

Distance-learning For Approximate Bayesian Computation To Model a Volcanic Eruption [article]

Lorenzo Pacchiardi, Pierre Kunzli, Marcel Schoengens, Bastien Chopard, Ritabrata Dutta
2019 arXiv   pre-print
However, informative summary statistics cannot be found for the considered model. We therefore develop a technique to learn a distance between model outputs based on deep metric-learning.  ...  Using ABC, which depends on many simulations from the considered model, we develop an inferential framework to learn parameters of a stochastic numerical simulator of volcanic eruption.  ...  Acknowledgements We thank CADMOS for providing computing resources at the Swiss Super Computing Center.  ... 
arXiv:1909.13118v1 fatcat:mhgovelncrb6hc7rc6xmm3ybgq

Three‐dimensional distribution of groundwater residence time metrics in the glaciated United States using metamodels trained on general numerical simulation models

J. J. Starn, L. J. Kauffman, C. S. Carlson, J. E. Reddy, M. N. Fienen
2021 Water Resources Research  
A similar approach was applied by J. J. Starn and Belitz (2018) to estimate one RTD for each of 30 numerically modeled small-scale watersheds.  ...  Fienen et al. (2018) used metamodeling trained on a large-scale numerical simulation model for the Michigan Basin created by Feinstein et al. (2010) ; however, a large-scale multi-layered numerical simulation  ... 
doi:10.1029/2020wr027335 fatcat:ctmi3hwrzjhwppwp3mllxa3z3m

Metric Learning for Simulation Analytics

Graham Laidler, Lucy E. Morgan, Barry L. Nelson, Nicos G. Pavlidis
2020 2020 Winter Simulation Conference (WSC)  
This model is built on the premise of a system-specific measure of similarity between observations of the state, which we inform via metric learning.  ...  In this paper, we take a simulation analytics view of output analysis, turning to machine learning methods to uncover key insights from the dynamic sample path.  ...  ACKNOWLEDGMENTS We gratefully acknowledge the support of the EPSRC funded EP/L015692/1 STOR-i Centre for Doctoral Training and the EPSRC NPIF fund EP/S515127/1.  ... 
doi:10.1109/wsc48552.2020.9383904 fatcat:afke77arkjb3zegdzpbiqqnwre

Reconstruction of Simulated Magnetic Resonance Fingerprinting Using Accelerated Distance Metric Learning

Elmira Yazdan, Sajjad Aghabozorgi Sahaf, Hamidreza Saligheh Rad
2020 Frontiers in biomedical technologies  
Conclusion: Our numerical results show that learning a distance metric of the undersampled training data accompanied by a compressed dictionary improves the accuracy of the MRF matching and overcomes the  ...  We propose using a distance metric learning method as a matching algorithm and a Singular Value Decomposition (SVD) to compress the dictionary, intending to promote the accuracy of MRF and expedite the  ...  We conducted numerical simulations to demonstrate the effectiveness of our framework.  ... 
doi:10.18502/fbt.v7i1.2720 fatcat:6cnlf5yugnc5vl4a3ksnp7sulm

Greedy metrics in orthogonal greedy learning [article]

Lin Xu, Shaobo Lin, Jinshan Zeng, Zongben Xu
2014 arXiv   pre-print
Furthermore, we reveals that such a greedy metric can bring an adaptive termination rule on the premise of maintaining the prominent learning performance of OGL.  ...  Orthogonal greedy learning (OGL) is a stepwise learning scheme that adds a new atom from a dictionary via the steepest gradient descent and build the estimator via orthogonal projecting the target function  ...  This implies that, at least for a certain learning task, SGD is not the unique metric for OGL.  ... 
arXiv:1411.3553v1 fatcat:iimwt466affszi4s7oeswcktme

A Data-driven Market Simulator for Small Data Environments [article]

Hans Bühler, Blanka Horvath, Terry Lyons, Imanol Perez Arribas, Ben Wood
2020 arXiv   pre-print
Though in this sense generative market simulation is model-free, the concrete modelling choices are nevertheless decisive for the features of the simulated paths.  ...  Finally, we also propose a suitable performance evaluation metric for financial time series and discuss some connections of our Market Generator to deep hedging.  ...  of numerical simulations of market paths.  ... 
arXiv:2006.14498v1 fatcat:imtmblo64jgr5gp2xfcbneac3y

Curved feature metrics in models of visual cortex

Norbert Mayer, J.Michael Herrmann, Theo Geisel
2002 Neurocomputing  
Whereas the former safely can be represented in a Euclidean space, the latter are shown to require a Riemannian metric in order to reach qualitatively similar stationary structures under a standard learning  ...  We show that the non-Euclidean framework allows for a tentative explanation of the presence of the so-called "pinwheels" in feature maps and compare maps obtained numerically in the at high-dimensional  ...  Later numerical results are presented for di erently curved (including at) feature space metrics together with data from a high-dimensional simulation.  ... 
doi:10.1016/s0925-2312(02)00417-4 fatcat:4o5gdzgufzbsffweympmroeali

EmbNum: Semantic labeling for numerical values with deep metric learning [article]

Phuc Nguyen, Khai Nguyen, Ryutaro Ichise, Hideaki Takeda
2018 arXiv   pre-print
To address these problems, we propose a neural numerical embedding model (EmbNum) to learn useful representation vectors for numerical attributes without prior assumptions on the distribution of data.  ...  Semantic labeling for numerical values is a task of assigning semantic labels to unknown numerical attributes.  ...  To learn the similarity metric for DSL, we followed their guideline that using logistic regression to train the similarity metrics with training samples are the pairs of numerical attributes.  ... 
arXiv:1807.01367v2 fatcat:o2ii5hcmdbgstkkqy67zwwx7ju

Computer Vision and Metrics Learning for Hypothesis Testing: An Application of Q-Q Plot for Normality Test [article]

Ke-Wei Huang and Mengke Qiao and Xuanqi Liu and Siyuan Liu and Mingxi Dai
2019 arXiv   pre-print
This paper proposes a new deep-learning method to construct test statistics by computer vision and metrics learning.  ...  Our proposed method integrates four components based on deep learning: an image representation learning component of a Q-Q plot, a dimension reduction component, a metrics learning component that best  ...  Next, we will briefly review the literature for image classification and similarity metrics learning by deep learning.  ... 
arXiv:1901.07851v2 fatcat:bj55j77jvzhtnn67ywyzuzb5li

How to cheat with metrics in single-image HDR reconstruction [article]

Gabriel Eilertsen, Saghi Hajisharif, Param Hanji, Apostolia Tsirikoglou, Rafal K. Mantiuk, Jonas Unger
2021 arXiv   pre-print
Single-image high dynamic range (SI-HDR) reconstruction has recently emerged as a problem well-suited for deep learning methods.  ...  Here, we reproduce a typical evaluation using existing as well as simulated SI-HDR methods to demonstrate how different aspects of the problem affect objective quality metrics.  ...  For results to be consistent between camera simulations, we expect the evaluation done with either CRF to produce very similar results, but this is not the case.  ... 
arXiv:2108.08713v1 fatcat:qwla4cvvebhhhpaeejrvowpnam

Distributed Semi-Supervised Metric Learning

Pengcheng Shen, Xin Du, Chunguang Li
2016 IEEE Access  
Over the last decade, many pairwise-constraint-based metric learning algorithms have been developed to automatically learn application-specific metrics from data under similarity/dissimilarity datapair  ...  Our simulation results show that the metrics learned by the proposed distributed algorithms are very close to that of the corresponding centralized method in most cases.  ...  useful information for the metric learning.  ... 
doi:10.1109/access.2016.2632158 fatcat:lkfd5rwwbngk3n66w2vxmcoln4

Deep Learning for Efficient Reconstruction of High-Resolution Turbulent DNS Data [article]

Pranshu Pant, Amir Barati Farimani
2021 arXiv   pre-print
Within the domain of Computational Fluid Dynamics, Direct Numerical Simulation (DNS) is used to obtain highly accurate numerical solutions for fluid flows.  ...  Large Eddy Simulation (LES) presents a more computationally efficient approach for solving fluid flows on lower-resolution (LR) grids but results in an overall reduction in solution fidelity.  ...  Introduction Direct Numerical Simulation (DNS) is a highly accurate but expensive method for computationally solving the Navier-Stokes equations.  ... 
arXiv:2010.11348v2 fatcat:may6m754qrcbto3nr3jgzg732i
« Previous Showing results 1 — 15 out of 111,367 results