Takeru Miyato

Affiliation

I am now working at Preferred Networks, Inc. as a researcher.
I was Master student in Integrated System Biology Labratory,
Department of System Science, Graduate school of Informatics, Kyoto University.

Research Interests

• Semi-supervised and unsupervised learning with neural networks
• Generative adversarial networks and implicit models
• Learning on extremely large distributed systems

Education

• M.S. Informatics – Graduate School of Informatics, Kyoto University, April 2014 - March 2016.
• B.S. Electronic Engineering – Kyoto University, April 2010 - March 2014.

Publications

1. Ken Nakanishi, Shin-ichi Maeda, Takeru Miyato and Daisuke Okanohara
   Neural Multi-scale Image Compression.
   ACCV, 2018. (accepted for oral presentation).
   [arXiv]
2. Takeru Miyato, Shin-ichi Maeda, Masanori Koyama and Shin Ishii
   Virtual Adversarial Training : A Regularization Method for Supervised and Semi-Supervised Learning.
   IEEE TPAMI, 2018. (extended version of the paper published at ICLR2016)
   [code (TF)] [code (Chainer)] [arXiv]
3. Takeru Miyato, Toshiki Kataoka, Masanori Koyama and Yuichi Yoshida
   Spectral Normalization for Generative Adversarial Networks.
   ICLR, 2018. (accepted for oral presentation)
   [code] [code (on ImageNet)] [OpenReview]
4. Takeru Miyato and Masanori Koyama
   cGANs with Projection Discriminator.
   ICLR, 2018.
   [code] [OpenReview]
5. Weihua Hu, Takeru Miyato, Seiya Tokui, Eiichi Matsumoto and Masashi Sugiyama
   Learning Discrete Representations via Information Maximizing Self Augmented Training.
   ICML, 2017.
   [code (by @weihua916)] [arXiv]
6. Takeru Miyato, Andrew M. Dai and Ian Goodfellow
   Adversarial Training Methods for Semi-Supervised Text Classification.
   ICLR, 2017.
   [code (Chainer, by @aonotas)] [code (TF)] [arXiv] [slide] [poster] [OpenReview]
7. Takeru Miyato, Daisuke Okanohara, Shin-ichi Maeda and Masanori Koyama
   Synthetic gradient methods with Virtual Forward-Backward Networks.
   Workshop at ICLR, 2017
   [poster] [OpenReview]
8. Takeru Miyato, Shin-ichi Maeda, Masanori Koyama, Ken Nakae and Shin Ishii
   Distributional Smoothing with Virtual Adversarial Training.
   ICLR, 2016.
   [code] [arXiv] [slide] [poster]

Pre-prints

1. Jiren Jin, Richard G. Calland, Takeru Miyato, Brian K. Vogel and Hideki Nakayama
   Parameter Reference Loss for Unsupervised Domain Adaptation.
   arXiv preprint arXiv:1711.07170, 2017.
   [arXiv]
2. Yuichi Yoshida and Takeru Miyato
   Spectral Norm Regularization for Improving the Generalizability of Deep Learning.
   arXiv preprint arXiv:1705.10941, 2017.
   [arXiv]

Working Experiences

• Preferred Networks Inc., Tokyo, Japan, 09/2016 - (now)
• Working as a full-time researcher.
• Advanced Telecommunications Research Institute International (ATR), Kyoto, Japan, 06/2016 - 03/2018
• Visiting researcher, 04/2017 - 03/2018
• Visiting research engineer, 09/2016 - 03/2017
• Full-time research engineer, 06/2016 - 09/2016
• Google Inc., Mountain View CA, 01/2016 - 05/2016.
• Software Engineer Intern (Research oriented) at Google Brain, under supervision of Dr. Andrew M. Dai and Dr. Ian Goodfellow.
  
• Suntex, Inc., Osaka, 02/2013 - 10/2013.
• Used C# to program the movement and the UI of an industrial bending machine.
• Jeyes, Inc., Osaka 04/2012 - 08/2013.
• Used OpenGL and Objective-C for the geometry processing in the development of iPhone and Android application for augmented reality(AR).

Teaching Experiences

• Information processing practice A, in Kyoto University, Japan, 10/2013-03/2014.
• Lecture on deep neural networks, in Kyoto University, Japan, 07/2014.

Skills

• Especially skilled in Theano, Tensorflow and Chainer, python based deep learning frameworks.

Hosted on GitHub Pages — Theme by orderedlist