Presentation 2020-07-16
A Study on Trainable ISTA using Auto Encoder as Shrinkage Function for Image Recovery
Kento Yokoyama, Satoshi Takabe, Tadashi Wadayama,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) ISTA (Iterative Shrinkage-Thresholding Algorithm) is one of the basic algorithms used in compressed sensing to estimate sparse signals from observed signals. Recently, techniques such as Trainable ISTA that combine deep learning with compressed sensing algorithms have achieved high signal recovery performance. In this paper, we propose the DAE-ISTA that uses a DAE (Denoising AutoEncoder) with pre-trained image features as the shrinkage function of ISTA for image recovery. Numerical simulations based on real data have revealed that DAE-ISTA improves the image recovery performance compared with an existing method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) DAE / ISTA / compressed sensing / deep learning
Paper # IT2020-13
Date of Issue 2020-07-09 (IT)

Conference Information
Committee IT
Conference Date 2020/7/16(1days)
Place (in Japanese) (See Japanese page)
Place (in English) Online
Topics (in Japanese) (See Japanese page)
Topics (in English) Freshman session, General
Chair Tadashi Wadayama(Nagoya Inst. of Tech.)
Vice Chair Tetsuya Kojima(Tokyo Kosen)
Secretary Tetsuya Kojima(Yamaguchi Univ.)
Assistant Takahiro Ohta(Senshu Univ.)

Paper Information
Registration To Technical Committee on Information Theory
Language JPN-ONLY
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Study on Trainable ISTA using Auto Encoder as Shrinkage Function for Image Recovery
Sub Title (in English)
Keyword(1) DAE
Keyword(2) ISTA
Keyword(3) compressed sensing
Keyword(4) deep learning
1st Author's Name Kento Yokoyama
1st Author's Affiliation Nagoya Institute of Technology(NIT)
2nd Author's Name Satoshi Takabe
2nd Author's Affiliation Nagoya Institute of Technology(NIT)
3rd Author's Name Tadashi Wadayama
3rd Author's Affiliation Nagoya Institute of Technology(NIT)
Date 2020-07-16
Paper # IT2020-13
Volume (vol) vol.120
Number (no) IT-105
Page pp.pp.13-18(IT),
#Pages 6
Date of Issue 2020-07-09 (IT)