Hoo Chang Shin

Research Scientist at NVIDIA

email: ken(dot)shin(at)beejion(dot)com
NIH IRP blog: i am intramural blog 
linkedin: profile

Research interests
machine learning - deep learning, unsupervised learning, reinforcement learning (in the era of big-data)
artificial intelligence - computer vision, natural language processing, robotics
data science and statistics - (finding the recent developments on probabilistic programming very interesting)
signal processing (the common math and fun), and integrated systems (it's nice to have it fit on something portable)
discovery and analysis of biomarkers - data science on large "-omics" data (genomics, radiomics, ...)

Personal interests
philosophy and Buddhism (why are we here? why do we live?), psychology (who am I? and so - Zen)
skin-scuba diving (a different world), mountain sports (where we live - sort of), tennis (also kind of Zen - overcoming self, and fun), and others
electrical guitar playing (I’m an educated electrical engineer) and digital music production (I do computer science nowadays)
business, economics, start-up, and etc.

2013: PhD at the Institute of Cancer Research, University of London, United Kingdom
 - Dissertation: Unsupervised Feature Learning for the Detection of Organs and Tumours in Multi-parametric Clinical Magnetic Resonance Images
2008: Diplom Ingineur at the Technical University of Munich, Germany
 - Master's thesis: Performance Analysis and FPGA-DSP Based Implementation of GNU Radio OFDM Transmitter (link)
2004: Bachelor of Science at the Sogang University, Seoul, Korea
 - Bachelor's thesis: Optimization of PID Controller Coefficients for a DC Motor using Evolutionary Algorithm (in Korean) (link)

May 2020 - Now: Research Scientist, NVIDIA, USA
March 2017 - May 2020: Solutions Architect, NVIDIA, USA
March 2014 - November 2016: Postdoctoral Visiting Fellow, National Institutes of Health, USA
May 2009 - August 2009: Internship at the Audi AG, Ingolstadt, Germany
October 2007 - October 2008: Internship and Student Researcher (master's thesis) at the BMW AG, Munich, Germany

Selected Publications

(For full list please see my Google Scholar page)


BioMegatron: Larger Biomedical Domain Language Model,
HC Shin, Y Zhang, E Bakhturina, R Puri, M Patwary, M Shoeybi, R Mani,
The 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020)
(arXiv) (NVIDIA blog-1, blog-2)


GANDALF: Generative Adversarial Networks with Discriminator-Adaptive Loss Fine-Tuning for Alzheimer’s Disease Diagnosis from MRI,
HC Shin, A Ihsani, Z Xu, S Mandava, ST Sreenivas, C Forster, J Cha,
Medical Image Computing and Computer Assisted Intervention (MICCAI) 2020


GANBERT: Generative Adversarial Networks with Bidirectional Encoder Representations from Transformers for MRI to PET synthesis,
HC Shin, A Ihsani, S Mandava, ST Sreenivas, C Forster, J Cha,
arXiv PrePrint, U.S. Patent application, 2020

Medical Image Synthesis for Data Augmentation and Anonymization using Generative Adversarial Networks,
HC Shin, N Tenenholtz, J Rogers, C Schwarz, M Senjem, J Gunter, K Andriole, M Michalski,
MICCAI Workshop on Simulation and Synthesis in Medical Imaging - SASHIMI 2018
(arXiv) (Code) (NVIDIA blog)

Learning to Read Chest X-Rays: Recurrent Neural Cascade Model for Automated Image Annotation,
HC. Shin, Kirk Roberts, Le Lu, Dina Demner-Fushman, Jianhua Yao, Ronald M. Summers,
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016
(arXiv) (Poster) (Code) (Data) (NVIDIA blog-1, blog-2)

Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning,
HC Shin, HR Roth, M Gao, L Lu, Z Xu, I Nogues, J Yao, D Mollura, RM Summers
IEEE transactions on medical imaging (TMI)
Vol. 35, Issue 5 (Special Issue on Deep Learning), pp. 1285-1298, Feb. 2016.

Interleaved Text/Image Deep Mining on a Very Large-Scale Radiology Database for Automated Image Interpretation,
HC. Shin, Le Lu, Lauren Kim, Ari Seff, Jianhua Yao, Ronald M. Summers,
Journal of Machine Learning Research (JMLR), 17(107):1-31, 2016.
This is an extension to the CVPR 2015 paper.
The new part in this paper is presented at the CVPR 2015 Language and Vision Workshop (link)


Interleaved Text/Image Deep Mining on a Very Large-Scale Radiology Database,
HC. Shin, Le Lu, Lauren Kim, Ari Seff, Jianhua Yao, Ronald M. Summers,
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015

Stacked Autoencoders for Unsupervised Feature Learning and Multiple Organ Detection 
in a Pilot Study Using 4D Patient Data, 
HC. Shin, M. Orton, D. Collins, S. Doran, M. Leach,
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI),
Vol. 35, Issue 8 (Special Issue on Deep Learning), pp. 1930-1943, Aug. 2013.

Hybrid Clustering and Logistic Regression for Multi-Modal Brain Tumor Segmentation, 
Hoo-Chang Shin,
Medical Image Computing and Computer Assisted Intervention (MICCAI) Workshops and Challenges, 2012.
- original approach: (PDF) (Poster)
- on-site challenge results: (link) 
 -- the k-means for edema detection and the supervised learning approach for tumor detection were replaced
    by an Bayesian-update process for both edema and tumor detection at the on-site challenge.
   (partly due to an unexpected power-cut in the Institute of Cancer Research during the challenge date in my institute, the algorithm 
     had to run on a laptop than on a powerful desktop, and partly on an interest, but could not be fully tested prior to the submission)

Autoencoder in Time-Series Analysis for Unsupervised Tissue Characterisation 
in a Large Unlabelled Medical Image Dataset, 
HC. Shin, M. Orton, D. Collins, S. Doran, M. Leach,
IEEE International Conference on Machine Learning and Applications (ICMLA), 2011.

Tunnel Entrance Recognition for Video-based Driver Assistance Systems, 
C. Claus, HC. Shin, W. Stechele,
International Conference on Systems, Signals and Image Processing (IWSSIP), 2006.
(PDF) (Project)

Unpublished Project

Region of Interest search engine (like Google) for medical images.

Book Chapter


  "Chapter 7: Organ Detection Using Unsupervised Deep Learning", of the book 
  Medical Image Recognition, Segmentation and Parsing: Machine Learning and Multiple Object Approaches
  Edited by S. K. Zhou, Elsevier, 2015.

Invited Talks

"Deep mining in text/image on a hospital scale PACS database: early findings"
 -- IEEE Computer Vision and Pattern Recognition (CVPR) 2015, Workshop on Medical Computer Vision (link)
 -- Brown Bag Lecture series (103rd, July 2015), Lister Hill National Center for Biomedical Communications, National Library of Medicne (link)
 -- Computer Science Special Seminar (July 2015), Department of Computer Science, Johns Hopkins University

"Interleaved Text/Image Deep Mining on a Very Large-Scale Radiology Database" 
 -- National Institutes of Health 2nd annual Pi Day event (link - March 14th, 2016), PiCo Lightning Talk (link) (video - from ~34 min.)

"Towards human-level understanding of medical images by learning from the big data of electronic health records"
 -- Seminar Lecture Series (April 2016), Computer Science Department, Johns Hopkins University
 -- Brown Bag Lecture series (June 2016), Lister Hill National Center for Biomedical Communications, National Library of Medicne (link)

Entrepreneurship Training

- FAES Advanced Studies in Technology Transfer, 2015 - 2016 (link)
- NovoEd Technology Entrepreneurship I, Fall 2013 (link)
 -- our team's project (link), background works (link)

languages - fluent in English, German, and Korean. Basic knowledge of Japanese.
MOOC - a fan, supporter and one of the early takers
 -- edX - CS188.1x Artificial Intelligence (with distinction)
 -- and some other classes of MIT OpenCourseWare, iTunesU, and many other open materials... (thank you all!)