Filters








115,517 Hits in 4.0 sec

Spatially Adaptive Computation Time for Residual Networks [article]

Michael Figurnov, Maxwell D. Collins, Yukun Zhu, Li Zhang, Jonathan Huang, Dmitry Vetrov, Ruslan Salakhutdinov
2017 arXiv   pre-print
This paper proposes a deep learning architecture based on Residual Network that dynamically adjusts the number of executed layers for the regions of the image.  ...  We present experimental results showing that this model improves the computational efficiency of Residual Networks on the challenging ImageNet classification and COCO object detection datasets.  ...  Spatially Adaptive Computation Time In this section, we present Spatially Adaptive Computation Time (SACT).  ... 
arXiv:1612.02297v2 fatcat:zilk5cgcpfgwpcxgfut3ik4bsy

Spatially Adaptive Computation Time for Residual Networks

Michael Figurnov, Maxwell D. Collins, Yukun Zhu, Li Zhang, Jonathan Huang, Dmitry Vetrov, Ruslan Salakhutdinov
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
This paper proposes a deep learning architecture based on Residual Network that dynamically adjusts the number of executed layers for the regions of the image.  ...  We present experimental results showing that this model improves the computational efficiency of Residual Networks on the challenging ImageNet classification and COCO object detection datasets.  ...  Spatially Adaptive Computation Time In this section, we present Spatially Adaptive Computation Time (SACT).  ... 
doi:10.1109/cvpr.2017.194 dblp:conf/cvpr/FigurnovCZZHVS17 fatcat:slbxpjvbw5hwvpyxiwqhhiebpu

IamNN: Iterative and Adaptive Mobile Neural Network for Efficient Image Classification [article]

Sam Leroux, Pavlo Molchanov, Pieter Simoens, Bart Dhoedt, Thomas Breuel, Jan Kautz
2018 arXiv   pre-print
In this work we use these properties to design a network based on a ResNet but with parameter sharing and with adaptive computation time.  ...  The resulting network is much smaller than the original network and can adapt the computational cost to the complexity of the input image.  ...  We combined these two ideas of parameter sharing and adaptive computation time to design a building block for accurate deep neural networks with small model size and adaptive computational cost.  ... 
arXiv:1804.10123v1 fatcat:vkls76zzmjar3mfmufnxo53epi

Probabilistic Adaptive Computation Time [article]

Michael Figurnov, Artem Sobolev, Dmitry Vetrov
2017 arXiv   pre-print
The recently proposed Adaptive Computation Time mechanism can be seen as an ad-hoc relaxation of this model.  ...  Evaluation on ResNet shows that our method matches the speed-accuracy trade-off of Adaptive Computation Time, while allowing for evaluation with a simple deterministic procedure that has a lower memory  ...  Application: Probabilistic Spatially Adaptive Computation Time for Residual Networks Residual network (ResNet) [14, 15] is a deep convolutional neural network architecture that has been successfully  ... 
arXiv:1712.00386v1 fatcat:kkk4wd64nffr3aladc6cfvqa5i

Deep Adaptive Inference Networks for Single Image Super-Resolution [article]

Ming Liu, Zhilu Zhang, Liya Hou, Wangmeng Zuo, Lei Zhang
2020 arXiv   pre-print
In this paper, we take a step forward to address this issue by leveraging the adaptive inference networks for deep SISR (AdaDSR).  ...  For most existing methods, the computational cost of each SISR model is irrelevant to local image content, hardware platform and application scenario.  ...  map for spatially adaptive inference.  ... 
arXiv:2004.03915v1 fatcat:notlxvsq3vhrtdofflwxuhoubq

Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference [article]

Thomas Verelst, Tinne Tuytelaars
2020 arXiv   pre-print
We introduce a residual block where a small gating branch learns which spatial positions should be evaluated.  ...  Modern convolutional neural networks apply the same operations on every pixel in an image. However, not all image regions are equally important.  ...  The closest work to ours is probably Spatially Adaptive Computation Time (SACT) [11] .  ... 
arXiv:1912.03203v2 fatcat:u6cbwfytvnbmpa2qvowhyfzldy

MPRNet: Multi-Path Residual Network for Lightweight Image Super Resolution [article]

Armin Mehri, Parichehr B.Ardakani, Angel D.Sappa
2020 arXiv   pre-print
Multi-Path Residual Network designs with a set of Residual concatenation Blocks stacked with Adaptive Residual Blocks: (i) to adaptively extract informative features and learn more expressive spatial context  ...  Lightweight super resolution networks have extremely importance for real-world applications.  ...  Later on, to address this drawback, FSRCNN [8] and ESPCN [35] have been proposed to reduce the large computational and run time cost by upsampling the features near to the output of the network.  ... 
arXiv:2011.04566v1 fatcat:a77ljhho45dt5d4ahkb7hmh63e

R-STAN: Residual Spatial-Temporal Attention Network for Action Recognition

Quanle Liu, Xiangjiu Che, Mei Bie
2019 IEEE Access  
To solve this problem, we propose residual spatial-temporal attention network (R-STAN), a feed-forward convolutional neural network using residual learning and spatial-temporal attention mechanism for  ...  Together with the specific characteristic of residual learning, we are able to construct a very deep network for learning spatial-temporal information in videos.  ...  These blocks guide the network to pay more attention to the distinctive time steps and spatial locations.  ... 
doi:10.1109/access.2019.2923651 fatcat:zcwxqutp4jaerbdeb7h5f6lgnu

Inferring incompressible two-phase flow fields from the interface motion using physics-informed neural networks

Aaron B. Buhendwa, Stefan Adami, Nikolaus A. Adams
2021 Machine Learning with Applications  
An effective way to distribute spatial training points to fit the interface, i.e. the volume fraction field, and the residual points is proposed.  ...  They proposed a new method to distribute the training points for the residual of the PDE (residual points).  ...  George Em Karniadakis of Brown University for stimulating discussions as part of a collaboration enabled by the Alexander von Humboldt foundation.  ... 
doi:10.1016/j.mlwa.2021.100029 fatcat:wvatvqgiqnap3fbs5yfnklwpd4

Densely Residual Network with Dual Attention for Hyperspectral Reconstruction from RGB Images

Lixia Wang, Aditya Sole, Jon Yngve Hardeberg
2022 Remote Sensing  
To this end, in this paper, based on multiple lightweight densely residual modules, we propose a densely residual network with dual attention (DRN-DA), which utilizes advanced attention and adaptive fusion  ...  strategy for more efficient feature correlation learning and more powerful feature extraction.  ...  At the end of the adaptive fusion stage, global residual learning is introduced to keep the network stable.  ... 
doi:10.3390/rs14133128 fatcat:ne6vqtlt2rb7hedm3jauxs22oe

Spatiotemporal empirical mode decomposition of resting-state fMRI signals: application to global signal regression [article]

Narges Moradi, Mehdy Dousty, Roberto Sotero
2019 bioRxiv   pre-print
Resting-state functional connectivity MRI (rs-fcMRI) is a common method for mapping functional brain networks.  ...  For each SIMF, brain connectivity matrices were computed by means of the Pearson correlation between TIMFs of different brain areas.  ...  Acknowledgements We are grateful to Doug Phillips for generous assistance in computational work on Compute Canada cluster and Jordan Chad for proofreading the manuscript.  ... 
doi:10.1101/556555 fatcat:omem2xfd35gz3pqsxs3h6ii5ma

Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference

Thomas Verelst, Tinne Tuytelaars
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
We introduce a residual block where a small gating branch learns which spatial positions should be evaluated.  ...  Modern convolutional neural networks apply the same operations on every pixel in an image. However, not all image regions are equally important.  ...  The closest work to ours is probably Spatially Adaptive Computation Time (SACT) [12] .  ... 
doi:10.1109/cvpr42600.2020.00239 dblp:conf/cvpr/VerelstT20 fatcat:uesc4pdzj5hrtdtbr5zd6ypubm

Spatiotemporal Empirical Mode Decomposition of Resting-State fMRI Signals: Application to Global Signal Regression

Narges Moradi, Mehdy Dousty, Roberto C. Sotero
2019 Frontiers in Neuroscience  
patterns for the default mode network and task-positive network.  ...  For each SIMF, functional connectivity matrices were computed by means of Pearson correlation between TIMFs of different brain areas.  ...  ACKNOWLEDGMENTS We are grateful to Doug Phillips for generous assistance in computational work on Compute Canada cluster and Jordan Chad for proofreading the manuscript.  ... 
doi:10.3389/fnins.2019.00736 pmid:31396032 pmcid:PMC6664052 fatcat:lwak7ljs2jg4zeojw5zdt4j64i

LogoNet: A Robust Layer-Aggregated Dual-Attention Anchorfree Logo Detection Framework with an Adversarial Domain Adaptation Approach

Rahul Kumar Jain, Taro Watasue, Tomohiro Nakagawa, Takahiro Sato, Yutaro Iwamoto, Xiang Ruan, Yen-Wei Chen
2021 Applied Sciences  
The proposed lightweight architecture significantly reduces the number of network parameters and improves the inference time to address the real-time performance while maintaining accuracy.  ...  It includes an hourglass-like top-down bottom-up feature extraction network, a spatial attention module and an anchorfree detection head similar to CenterNet.  ...  In future, we will discover more attention-and domain adaptation-based mechanisms including transformer [40] and lightweight compact network for logo detection in real-time.  ... 
doi:10.3390/app11209622 fatcat:kmwevguo7vdjvl64z7v4ydw3n4

Adaptive Regularization via Residual Smoothing in Deep Learning Optimization [article]

Junghee Cho, Junseok Kwon, Byung-Woo Hong
2019 arXiv   pre-print
The degree of regularization at each element in the target space of the neural network architecture is determined based on the residual at each optimization iteration in an adaptive way.  ...  Our adaptive regularization algorithm is designed to apply a diffusion process driven by the heat equation with spatially varying diffusivity depending on the probability density function following a certain  ...  In addition to the computational efficiency, it is desired to consider the relative magnitude of residual in its spatial domain.  ... 
arXiv:1907.09750v2 fatcat:2futs44xvnbzhgxlwbywpwkb3m
« Previous Showing results 1 — 15 out of 115,517 results