Filters








400 Hits in 6.1 sec

Improving Convolutional Neural Networks Via Conservative Field Regularisation and Integration [article]

Dominique Beaini, Sofiane Achiche, Maxime Raison
2020 arXiv   pre-print
Current research in convolutional neural networks (CNN) focuses mainly on changing the architecture of the networks, optimizing the hyper-parameters and improving the gradient descent.  ...  field by forcing it to be conservative.  ...  Introduction Since the year 2015, the convolutional neural networks (CNN) rose quickly to become the best machine learning technique used to solve many computer vision problems such as classification  ... 
arXiv:2003.05182v1 fatcat:mdruxuggzzbkzlmrxtkxryzomu

Structure preserving deep learning [article]

Elena Celledoni, Matthias J. Ehrhardt, Christian Etmann, Robert I McLachlan, Brynjulf Owren, Carola-Bibiane Schönlieb, Ferdia Sherry
2020 arXiv   pre-print
In this article, we review a number of these directions: some deep neural networks can be understood as discretisations of dynamical systems, neural networks can be designed to have desirable properties  ...  such as invertibility or group equivariance, and new algorithmic frameworks based on conformal Hamiltonian systems and Riemannian manifolds to solve the optimisation problems have been proposed.  ...  and vision (2017) and Geometry, compatibility and structure preservation in computational differential equations (2019) where work on this paper was undertaken, EPSRC grant EP/K032208/1.  ... 
arXiv:2006.03364v1 fatcat:dy5t5w4gsfeavl72e3oqllnlqe

Structure-preserving deep learning

E. CELLEDONI, M. J. EHRHARDT, C. ETMANN, R. I. MCLACHLAN, B. OWREN, C.-B. SCHONLIEB, F. SHERRY
2021 European journal of applied mathematics  
In this article, we review a number of these directions: some deep neural networks can be understood as discretisations of dynamical systems, neural networks can be designed to have desirable properties  ...  such as invertibility or group equivariance and new algorithmic frameworks based on conformal Hamiltonian systems and Riemannian manifolds to solve the optimisation problems have been proposed.  ...  MJE acknowledges support from the EPSRC grants EP/S026045/1 and EP/T026693/1, the Faraday Institution via EP/T007745/  ... 
doi:10.1017/s0956792521000139 fatcat:zhynqzjuorbcbmz62nm6t44fzi

End-to-End Conditional GAN-based Architectures for Image Colourisation [article]

Marc Górriz, Marta Mrak, Alan F. Smeaton, Noel E. O'Connor
2019 arXiv   pre-print
In this work recent advances in conditional adversarial networks are investigated to develop an end-to-end architecture based on Convolutional Neural Networks (CNNs) to directly map realistic colours to  ...  In particular, the method uses Generative Adversarial Neural Networks (GANs) and focuses on improvement of training stability to enable better generalisation in large multi-class image datasets.  ...  end-to-end Convolutional Neural Network architectures.  ... 
arXiv:1908.09873v2 fatcat:3ogj3lcypbaijpxous7wvibswm

Physics-Guided Deep Learning for Dynamical Systems: A Survey [article]

Rui Wang, Rose Yu
2022 arXiv   pre-print
In this paper, we provide a structured overview of existing methodologies of integrating prior physical knowledge or physics-based modeling into DL, with a special emphasis on learning dynamical systems  ...  Thus, the study of physics-guided DL emerged and has gained great progress.  ...  neural networks that respect conservation laws.  ... 
arXiv:2107.01272v5 fatcat:k6hhdt6csnfebgkzrpuoeqkwzi

The Hidden Uncertainty in a Neural Networks Activations [article]

Janis Postels, Hermann Blum, Yannick Strümpler, Cesar Cadena, Roland Siegwart, Luc Van Gool, Federico Tombari
2021 arXiv   pre-print
While our approach does not require modifying the training process, we follow prior work and experiment with an additional regularising loss that increases the information in the latent representations  ...  The distribution of a neural network's latent representations has been successfully used to detect out-of-distribution (OOD) data.  ...  A practical bayesian framework for backprop- agation networks. Neural computation, 4(3):448-472, 1992. Malinin, A. and Gales, M. Predictive uncertainty estimation via prior networks.  ... 
arXiv:2012.03082v2 fatcat:2txmq45dhbaytgoc7vn6dyvwuu

A weighted least squares optimisation strategy for medical image super resolution via multiscale convolutional neural networks for healthcare applications

Bhawna Goyal, Dawa Chyophel Lepcha, Ayush Dogra, Shui-Hua Wang
2021 Complex & Intelligent Systems  
In this study, an effective medical super-resolution approach based on weighted least squares optimisation via multiscale convolutional neural networks (CNNs) has been proposed for lesion localisation.  ...  The trained multiple convolutional neural networks characterize medical images in many directions and multiscale frequency bands, and thus facilitate image restoration subject to increased number of variations  ...  Our multiscale convolutional neural networks improve previous method by training multiple convolutional neural networks to characterize wavelet multiscale representations.  ... 
doi:10.1007/s40747-021-00465-z fatcat:c2t3qd4yjvhhfgggi5d6pulfx4

Neural physical engines for inferring the halo mass distribution function [article]

Tom Charnock, Guilhem Lavaux, Benjamin D. Wandelt, Supranta Sarma Boruah, Jens Jasche, Michael J. Hudson
2019 arXiv   pre-print
The method is based on the Bayesian analysis of simple, physically motivated, neural network-like architectures, which we denote as neural physical engines, and neural density estimation.  ...  Here we demonstrate a method for determining the halo mass distribution function by learning the tracer bias between density fields and halo catalogues using a neural bias model.  ...  This work has been done within the activities of the Domaine d'Intérêt Majeur (DIM) "Astrophysique et Conditions d'Apparition de la Vie" (ACAV), and received financial support from RégionÎle-de-France.  ... 
arXiv:1909.06379v1 fatcat:x54bkl6pejcwxgpucnstfzsu4a

Data augmentation and image understanding [article]

Alex Hernandez-Garcia
2020 arXiv   pre-print
Throughout this dissertation, I use these insights to analyse data augmentation as a particularly useful inductive bias, a more effective regularisation method for artificial neural networks, and as the  ...  framework to analyse and improve the invariance of vision models to perceptually plausible transformations.  ...  of current convolutional neural networks-directly inspired by the findings by Hubel & Wiesel (1959) about the receptive fields of neurons in the cat's visual cortex.  ... 
arXiv:2012.14185v1 fatcat:qcip4vstzvbxzo4qevek5marrm

Computer Audition for Continuous Rainforest Occupancy Monitoring: The Case of Bornean Gibbons' Call Detection

Panagiotis Tzirakis, Alexander Shiarella, Robert Ewers, Björn W. Schuller
2020 Interspeech 2020  
Autonomous recording units (ARUs) enable auditory data collection over a wider area, and can provide improved consistency over traditional sampling methods.  ...  In this paper, we address the divide between academic machine learning research on animal vocalisation classifiers, and their application to conservation efforts.  ...  Due to the high number of parameters our model contains, we use batch normalisation [27] as regularisation after each convolution layer. Recurrent neural network.  ... 
doi:10.21437/interspeech.2020-2655 dblp:conf/interspeech/TzirakisSES20 fatcat:plnc2disbjgc3pqdq2sgxuufqy

Towards Physics-informed Deep Learning for Turbulent Flow Prediction [article]

Rui Wang, Karthik Kashinath, Mustafa Mustafa, Adrian Albert, Rose Yu
2020 arXiv   pre-print
Most importantly, our method predicts physical fields that obey desirable physical characteristics, such as conservation of mass, whilst faithfully emulating the turbulent kinetic energy field and spectrum  ...  In this paper, we aim to predict turbulent flow by learning its highly nonlinear dynamics from spatiotemporal velocity fields of large-scale fluid flow simulations of relevance to turbulence modeling and  ...  We thank Jared Dunnmon and Maziar Raissi for helpful discussions. We also thank Dragos Bogdan Chirila for providing the turbulent flow data.  ... 
arXiv:1911.08655v4 fatcat:qnjvnynhbja6rjhcmrx77r3jfu

Hierarchical Attentive Recurrent Tracking [article]

Adam R. Kosiorek, Alex Bewley, Ingmar Posner
2017 arXiv   pre-print
To improve training convergence, we augment the loss function with terms for a number of auxiliary tasks relevant for tracking.  ...  Inspired by how the human visual cortex employs spatial attention and separate "where" and "what" processing pathways to actively suppress irrelevant visual features, this work develops a hierarchical  ...  Acknowledgements We would like to thank Oiwi Parker Jones and Martin Engelcke for discussions and valuable insights and Neil Dhir for his help with editing the paper.  ... 
arXiv:1706.09262v2 fatcat:gvhhwavwjnek5ijpddesyjeipe

Using transfer learning to detect galaxy mergers

Sandro Ackermann, Kevin Schawinski, Ce Zhang, Anna K Weigel, M Dennis Turp
2018 Monthly notices of the Royal Astronomical Society  
We investigate the use of deep convolutional neural networks (deep CNNs) for automatic visual detection of galaxy mergers.  ...  Transfer learning on our full training set leads to a lowered error rate from 0.038 ± 1 down to 0.032 ± 1, a relative improvement of 15 our method, and comparing with an already existing, manually created  ...  Recently, with the advent of large labeled datasets and cheap computational resources, Convolutional Neural Networks (CNNs) have achieved a performance level that represents a significant improvement over  ... 
doi:10.1093/mnras/sty1398 fatcat:yhuueyxv75glbmj3u4dh2bzx4a

Learning Wave Propagation with Attention-Based Convolutional Recurrent Autoencoder Net [article]

Indu Kant Deo, Rajeev Jaiman
2022 arXiv   pre-print
The proposed deep neural network architecture relies on the attention-based recurrent neural network with long short-term memory cells.  ...  Denoising autoencoder further reduces the mean squared error of prediction and improves the generalization capability in the parameter space.  ...  ACKNOWLEDGEMENTS The authors would like to acknowledge the funding support from the University of British Columbia (UBC) and the Natural Sciences and Engineering Research Council of Canada (NSERC).  ... 
arXiv:2201.06628v3 fatcat:gr74fjdqrfezfdhm5bt4ajlhjy

Transient-optimised real-bogus classification with Bayesian Convolutional Neural Networks – sifting the GOTO candidate stream [article]

T. L. Killestein, J. Lyman, D. Steeghs, K. Ackley, M. J. Dyer, K. Ulaczyk, R. Cutter, Y.-L. Mong, D. K. Galloway, V. Dhillon, P. O'Brien, G. Ramsay (+36 others)
2021 arXiv   pre-print
In this paper, we present a new real-bogus classifier based on a Bayesian convolutional neural network that provides nuanced, uncertainty-aware classification of transient candidates in difference imaging  ...  , and demonstrate its application to the datastream from the GOTO wide-field optical survey.  ...  DS, KU, BG and JDL acknowledge support from the STFC via grants ST/T007184/1, ST/T003103/1 and ST/P000495/1. JDL acknowledges support from a UK Research and Innovation Fellowship (MR/T020784/1).  ... 
arXiv:2102.09892v1 fatcat:hwdg5xootrdjnljvp73w5yz33a
« Previous Showing results 1 — 15 out of 400 results