Filters








9,270 Hits in 4.5 sec

Deep-neural-network based sinogram synthesis for sparse-view CT image reconstruction

Hoyeon Lee, Jongha Lee, Hyeongseok Kim, Byungchul Cho, Seungryong Cho
2018 IEEE Transactions on Radiation and Plasma Medical Sciences  
In this work, we introduce a deep-neural-network-enabled sinogram synthesis method for sparse-view CT, and show its outperformance to the existing interpolation methods and also to the iterative image  ...  Interpolation methods have also been utilized to fill the missing data in the sinogram of sparse-view CT thus providing synthetically full data for analytic image reconstruction.  ...  Greatly inspired by the recent progresses of machine learning techniques, we propose in this work to use a deep-neural-network for synthesizing the missing data in the sparse-view sinograms.  ... 
doi:10.1109/trpms.2018.2867611 fatcat:7mb5jcn2jfacjlhka7e3pwgiu4

Deep Learning with Inaccurate Training Data for Image Restoration [article]

Bolin Liu, Xiao Shu, Xiaolin Wu
2018 arXiv   pre-print
In such cases, one is forced to use synthesized paired data to train the deep convolutional neural network (DCNN).  ...  In many applications of deep learning, particularly those in image restoration, it is either very difficult, prohibitively expensive, or outright impossible to obtain paired training data precisely as  ...  Experimental Results Shown in Conclusion In many applications of deep learning, we have to resort to synthesized paired data to train the DCNN, due to the lack of real data.  ... 
arXiv:1811.07268v1 fatcat:2qxi4htovjawdefcmnjccfbdz4

DeepMotifSyn: a deep learning approach to synthesize heterodimeric DNA motifs

Jiecong Lin, Lei Huang, Xingjian Chen, Shixiong Zhang, Ka-Chun Wong
2021 Briefings in Bioinformatics  
The generator is a U-Net-based neural network that can synthesize heterodimeric motifs from aligned motif pairs.  ...  We introduce DeepMotifSyn, a deep learning-based tool for synthesizing heterodimeric motifs from monomeric motif pairs.  ...  Acknowledgments We gratefully acknowledge the support of NVIDIA Corporation with the Titan XP GPU for this research.  ... 
doi:10.1093/bib/bbab334 pmid:34524404 fatcat:hcrlhzvdfjelfbvrzny5buhtiy

Strategies for Training Deep Learning Models in Medical Domains with Small Reference Datasets

Gerald A. Zwettler, David R. Holmes III, Werner Backfrieder
2020 Journal of WSCG  
While training on 2,200 real images only leads to accuracy JI=88.75, the enrichment with 2,200 additional images synthesized from a GAN trained on 5,000 datasets only leads to an increase up to JI=92.02  ...  In this work the applicability of a U-net cascade for training on a very low amount of abdominal MRI datasets of the parenchyma is evaluated and strategies to compensate for the lack of training data are  ...  METHODOLOGY The Deep Learning Network to be used in this paper for validation on low or synthesized data is a U-net [Ron15] applied as cascade with combining axial, sagittal and coronal views [Zwe20  ... 
doi:10.24132/jwscg.2020.28.5 fatcat:jle43tluxzflri5wnlgt54yska

Bespoke Neural Networks for Score-Informed Source Separation [article]

Ethan Manilow, Bryan Pardo
2020 arXiv   pre-print
This lets us create a labeled training set to train a network on the specific bespoke task.  ...  Given an unaligned MIDI transcription for a target instrument from an input mixture, we synthesize new mixtures from the midi transcription that sound similar to the mixture to be separated.  ...  Using the known synthesized source as the ground truth source label, we train a small neural network to "overfit" to these local augmented examples as surrogate to the actual mixture.  ... 
arXiv:2009.13729v1 fatcat:mvpyxjkipzcz5asl2jrrlwvdc4

Predictive and generative neural networks for object functionality

Ruizhen Hu, Zihao Yan, Jingwen Zhang, Oliver Van Kaick, Ariel Shamir, Hao Zhang, Hui Huang
2018 ACM Transactions on Graphics  
We develop predictive and generative deep convolutional neural networks to replicate this feat.  ...  interaction contexts. fSIM-NET is complemented by a generative network (iGEN-NET) and a segmentation network (iSEG-NET). iGEN-NET takes a single voxelized 3D object with a functionality label and synthesizes  ...  To evaluate the diversity of the synthesized output, we perform a comparison of the variation in the training data compared to the variation in the synthesized data.  ... 
doi:10.1145/3197517.3201287 fatcat:so7cbtezyfecngw6tnwjxi2shm

Sparse Pixel Training of Convolutional Neural Networks for Land Cover Classification

Noureldin Laban, Bassam Abdellatif, Hala M. Ebeid, Howida A. Shedeed, Mohamed F. Tolba
2021 IEEE Access  
According to SegNet architecture, we have used the traditional one with our synthesized image dataset used with U-Net. SegNet was train using 700 epochs.  ...  The second perspective is dealing with data where large datasets are built to be used in the training process of different deep learning architectures.  ... 
doi:10.1109/access.2021.3069882 fatcat:gq5u4diqnfcq3g6q4kyx5ofb7m

The Effectiveness of Data Augmentation for Bone Suppression in Chest Radiograph using Convolutional Neural Network

Ren G, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, China, Lam S-K, Ni R, Yang D, Qin J, Cai J, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, China, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, China, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, China, School of Nursing, The Hong Kong Polytechnic University, Hong Kong SAR, China, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong SAR, China
2021 Austin Journal of Cancer and Clinical Research  
However, data scarcity has long been considered as the prime culprit of developing Convolutional Neural Networks (CNNs) models for the task of bone suppression.  ...  Two CNN models (U-Net and Generative Adversarial Network (GAN)) were adapted to explore the effectiveness of various data augmentation techniques for bone signal suppression in the chest radiograph.  ...  [13] , frequency-specific deep neural network convolution [14] , to name a few.  ... 
doi:10.26420/austinjcancerclinres.2021.1095 fatcat:ybffi7za6fh2xf6w6xgvfw37nq

READ: Large-Scale Neural Scene Rendering for Autonomous Driving [article]

Zhuopeng Li, Lu Li, Zeyu Ma, Ping Zhang, Junbo Chen, Jianke Zhu
2022 arXiv   pre-print
In this paper, a large-scale neural rendering method is proposed to synthesize the autonomous driving scene (READ), which makes it possible to synthesize large-scale driving scenarios on a PC through a  ...  In order to represent driving scenarios, we propose an ω rendering network to learn neural descriptors from sparse point clouds.  ...  synthesize the scene with few observations, resulting in blur.  ... 
arXiv:2205.05509v1 fatcat:nasvfsqyjbekldqucynvc5llay

The Effectiveness of Instance Normalization: a Strong Baseline for Single Image Dehazing [article]

Zheng Xu, Xitong Yang, Xue Li, Xiaoshuai Sun
2018 arXiv   pre-print
We propose a novel deep neural network architecture for the challenging problem of single image dehazing, which aims to recover the clear image from a degraded hazy image.  ...  We adopt the convolutional layers of the pre-trained VGG network as encoder to exploit the representation power of deep features, and demonstrate the effectiveness of instance normalization for image dehazing  ...  In contrast, training deep neural networks with large-scale data has made significant progress and achieved state-of-the-art performance in many vision tasks [21, 35, 16] .  ... 
arXiv:1805.03305v1 fatcat:22avqamhnzcfvedq3kepoq2yde

Musical Audio Synthesis Using Autoencoding Neural Nets

Andy M. Sarroff, Michael Casey
2014 Proceedings of the SMC Conferences  
(Abstract to follow)  ...  We note that this is often the order of events for training a deep neural network.  ...  There are several disadvantages to using FFTs as the low level training data; these are discussed later.  ... 
doi:10.5281/zenodo.850877 fatcat:7fc5juwmdrd2ffb74ua3e64evq

Deep neural networks for automated detection of marine mammal species

Yu Shiu, K. J. Palmer, Marie A. Roch, Erica Fleishman, Xiaobai Liu, Eva-Marie Nosal, Tyler Helble, Danielle Cholewiak, Douglas Gillespie, Holger Klinck
2020 Scientific Reports  
We demonstrate that a deep neural network trained with recordings from a single geographic region recorded over a span of days is capable of generalizing well to data from multiple years and across the  ...  The deep neural networks we developed are relatively easy to implement with existing software, and may provide new insights applicable to the conservation of endangered species.  ...  Pelkie for helping with data analysis; F. Channell, C. Tessaglia-Hymes, and D. Jaskula for deploying and retrieving MARUs, and the DCLDE 2013 organizing committee.  ... 
doi:10.1038/s41598-020-57549-y pmid:31953462 pmcid:PMC6969184 fatcat:3l3cupmq3fet5h53jylsb6p3m4

Using Deep Learning Techniques and Inferential Speech Statistics for AI Synthesised Speech Recognition [article]

Arun Kumar Singh
2021 arXiv   pre-print
The recent developments in technology have re-warded us with amazing audio synthesis models like TACOTRON and WAVENETS.  ...  The temporal dependencies present in AI synthesized speech are exploited using Bidirectional RNN and CNN.  ...  Most of the synthesised speeches are generated using powerful AI algorithms and training of deep neural networks.  ... 
arXiv:2107.11412v1 fatcat:q3sgzfxclzgxzf32menjpovrj4

Towards computer-aided severity assessment via deep neural networks for geographic and opacity extent scoring of SARS-CoV-2 chest X-rays

A. Wong, Z. Q. Lin, L. Wang, A. G. Chung, B. Shen, A. Abbasi, M. Hoshmand-Kochi, T. Q. Duong
2021 Scientific Reports  
The deep neural networks used in this study, which we name COVID-Net S, are based on a COVID-Net network architecture. 100 versions of the network were independently learned (50 to perform geographic extent  ...  Chest x-rays (CXRs) are often used to assess SARS-CoV-2 severity, with two important assessment metrics being extent of lung involvement and degree of opacity.  ...  Acknowledgements We would like to thank Natural Sciences and Engineering Research Council of Canada (NSERC), the Canada Research Chairs program, CIFAR, DarwinAI Corp., Nvidia Corp., Hewlett Packard Enterprise  ... 
doi:10.1038/s41598-021-88538-4 pmid:33927239 fatcat:2vs522uvenfkzkkko5qygtbfc4

You Only Look Yourself: Unsupervised and Untrained Single Image Dehazing Neural Network [article]

Boyun Li, Yuanbiao Gou, Shuhang Gu, Jerry Zitao Liu, Joey Tianyi Zhou, Xi Peng
2020 arXiv   pre-print
In this paper, we study two challenging and less-touched problems in single image dehazing, namely, how to make deep learning achieve image dehazing without training on the ground-truth clean image (unsupervised  ...  Thanks to the unsupervised and untrained characteristics of YOLY, our method bypasses the conventional training paradigm of deep models on hazy-clean pairs or a large scale dataset, thus avoids the labor-intensive  ...  However, it has to use a corrupted image set with the same noise distribution to train the neural network.  ... 
arXiv:2006.16829v1 fatcat:wokohyplyrd77ouujz3fkohjgm
« Previous Showing results 1 — 15 out of 9,270 results