TensorView: Visualizing the training of convolutional neural network using paraview

Xinyu Chen; Qiang Guan; Xin Liang; Li Ta Lo; Simon Su; Trilce Estrada; James Ahrens

doi:10.1145/3154842.3154846

TensorView: Visualizing the training of convolutional neural network using paraview

Xinyu Chen
, Qiang Guan
, Xin Liang
, Li Ta Lo
, Simon Su
, Trilce Estrada
, James Ahrens

Producción científica: Conference contribution › revisión exhaustiva

3 Citas (Scopus)

Resumen

Convolutional Neural Networks(CNNs) have been widely used in visual recognition tasks recently. Previous works visualize learning features at different layers to help people to understand how CNNs learn visual recognition tasks. However they only provide qualitative description and do not help to accelerate the training process. We present TensorView to enable Paraview to visualize the evolution of CNNs. TensorView provides both qualitative and quantitative visualization that help understand the learning procedure, tune the learning parameters, direct merging and pruning of neural networks.

Idioma original	English
Título de la publicación alojada	DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017
Páginas	11-16
Número de páginas	6
ISBN (versión digital)	9781450351690
DOI	https://doi.org/10.1145/3154842.3154846
Estado	Published - dic 11 2017
Evento	1st Workshop on Distributed Infrastructures for Deep Learning, DIDL 2017 - Las Vegas, United States Duración: dic 11 2017 → dic 15 2017

Serie de la publicación

Nombre	DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017

Conference

Conference	1st Workshop on Distributed Infrastructures for Deep Learning, DIDL 2017
País/Territorio	United States
Ciudad	Las Vegas
Período	12/11/17 → 12/15/17

Nota bibliográfica

Publisher Copyright:
© 2017 Copyright held by the owner/author(s).

ASJC Scopus subject areas

Computer Networks and Communications
Hardware and Architecture
Information Systems
Software

Acceder al documento

10.1145/3154842.3154846

Otros archivos y enlaces

Citar esto

Chen, X., Guan, Q., Liang, X., Lo, L. T., Su, S., Estrada, T., & Ahrens, J. (2017). TensorView: Visualizing the training of convolutional neural network using paraview. En DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017 (pp. 11-16). (DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017). https://doi.org/10.1145/3154842.3154846

Chen, Xinyu ; Guan, Qiang ; Liang, Xin et al. / TensorView : Visualizing the training of convolutional neural network using paraview. DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017. 2017. pp. 11-16 (DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017).

@inproceedings{4639ce629cb441b0a85e88fd1c7d9a9a,

title = "TensorView: Visualizing the training of convolutional neural network using paraview",

abstract = "Convolutional Neural Networks(CNNs) have been widely used in visual recognition tasks recently. Previous works visualize learning features at different layers to help people to understand how CNNs learn visual recognition tasks. However they only provide qualitative description and do not help to accelerate the training process. We present TensorView to enable Paraview to visualize the evolution of CNNs. TensorView provides both qualitative and quantitative visualization that help understand the learning procedure, tune the learning parameters, direct merging and pruning of neural networks.",

keywords = "Convolutional networks, Paraview, Visualization",

author = "Xinyu Chen and Qiang Guan and Xin Liang and Lo, \{Li Ta\} and Simon Su and Trilce Estrada and James Ahrens",

note = "Publisher Copyright: {\textcopyright} 2017 Copyright held by the owner/author(s).; 1st Workshop on Distributed Infrastructures for Deep Learning, DIDL 2017 ; Conference date: 11-12-2017 Through 15-12-2017",

year = "2017",

month = dec,

day = "11",

doi = "10.1145/3154842.3154846",

language = "English",

series = "DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017",

pages = "11--16",

booktitle = "DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017",

}

Chen, X, Guan, Q, Liang, X, Lo, LT, Su, S, Estrada, T & Ahrens, J 2017, TensorView: Visualizing the training of convolutional neural network using paraview. En DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017. DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017, pp. 11-16, 1st Workshop on Distributed Infrastructures for Deep Learning, DIDL 2017, Las Vegas, United States, 12/11/17. https://doi.org/10.1145/3154842.3154846

TensorView: Visualizing the training of convolutional neural network using paraview. / Chen, Xinyu; Guan, Qiang; Liang, Xin et al.
DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017. 2017. p. 11-16 (DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017).

Producción científica: Conference contribution › revisión exhaustiva

TY - GEN

T1 - TensorView

T2 - 1st Workshop on Distributed Infrastructures for Deep Learning, DIDL 2017

AU - Chen, Xinyu

AU - Guan, Qiang

AU - Liang, Xin

AU - Lo, Li Ta

AU - Su, Simon

AU - Estrada, Trilce

AU - Ahrens, James

PY - 2017/12/11

Y1 - 2017/12/11

N2 - Convolutional Neural Networks(CNNs) have been widely used in visual recognition tasks recently. Previous works visualize learning features at different layers to help people to understand how CNNs learn visual recognition tasks. However they only provide qualitative description and do not help to accelerate the training process. We present TensorView to enable Paraview to visualize the evolution of CNNs. TensorView provides both qualitative and quantitative visualization that help understand the learning procedure, tune the learning parameters, direct merging and pruning of neural networks.

AB - Convolutional Neural Networks(CNNs) have been widely used in visual recognition tasks recently. Previous works visualize learning features at different layers to help people to understand how CNNs learn visual recognition tasks. However they only provide qualitative description and do not help to accelerate the training process. We present TensorView to enable Paraview to visualize the evolution of CNNs. TensorView provides both qualitative and quantitative visualization that help understand the learning procedure, tune the learning parameters, direct merging and pruning of neural networks.

KW - Convolutional networks

KW - Paraview

KW - Visualization

UR - https://www.scopus.com/pages/publications/85041179496

UR - https://www.scopus.com/pages/publications/85041179496#tab=citedBy

U2 - 10.1145/3154842.3154846

DO - 10.1145/3154842.3154846

M3 - Conference contribution

AN - SCOPUS:85041179496

T3 - DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017

SP - 11

EP - 16

BT - DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017

Y2 - 11 December 2017 through 15 December 2017

ER -

Chen X, Guan Q, Liang X, Lo LT, Su S, Estrada T et al. TensorView: Visualizing the training of convolutional neural network using paraview. En DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017. 2017. p. 11-16. (DIDL 2017 - Proceedings of the 1st Workshop on Distributed Infrastructures for Deep Learning, Part of Middleware 2017). doi: 10.1145/3154842.3154846

TensorView: Visualizing the training of convolutional neural network using paraview

Resumen

Serie de la publicación

Conference

Nota bibliográfica

ASJC Scopus subject areas

Acceder al documento

Otros archivos y enlaces

Huella

Citar esto