증명사진.jpg

Junho Kim (1993.01.12)

Kyungbok High School, Seoul (2011.02)

B.S. in Chung-Ang University, Seoul (2015.02)
 - Major in Computer Science and Engineering
 - Double Major in Mathematics

M.S. in Chung-Ang University Graduate School, Seoul (2017.02)

 - Electronic Commerce & Internet Application Lab

 - Natural Language Processing Research

https://www.facebook.com/taki0112

https://github.com/taki0112

https://www.linkedin.com/in/taki0112

[email protected]


In now, resume -> http://bit.ly/jhkim_ai

Technical Skills


Programming

Strong : Tensorflow, Python, Java

Knowledgeable : C, C++

Knowledge

Deep Learning, Machine Learning

Natural Language Processing

Generative Adversarial Network

Awards


E-sports Competition

Organized by Korea e-Sports Association & Shinhan Bank, 2011

Representative of Chung-Ang University
Runner-up

Human ICT Idea Competition

Organized by Chung-Ang University, 2014

Coupon exchange mobile application for private cafe regeneration, Excellence Prize


Paper

Differential equation solution using numerical method and error minimization method of approximate solution (B.S. Thesis, 2014.12)

In this paper, I focus on the methodology that minimizes the error of the approximation error of the solution of the differential equation which plays an important role not only in computer engineering but also in engineering, physics and economics.

The online community based decision making support system for mitigating biased decision making (AIP Conference Proceedings, 2016.11)

I propose a decision support system that can support users' decision making by classifying the tendency (progress and remuneration) of opinion using opinion data of online group and classifying positive and negative by performing emotional analysis on opinion

Identifying the intent of documents by utilizing core passages extracted from short texts (M.S. Thesis, 2017.02)

I propose a method to calculate the similarity between documents based on extracting key phrases representing "intention" from short-length documents (eg, articles from microblogging services). This shows 80% performance improvement over the existing method. I have applied for a patent of this paper, and I am currently in the process of journalizing this paper.

Patent

Document similarity analysis apparatus and method (2017.03.29)

Apparatus and method for constructing dynamic term identification system of user-generated contents (2017.03.29)


Article

DeepLearning :: Summary :: [link]

CNN-Architecture :: Summary :: [link]

Activities


Twenty Network

2014.09 ~ 2015.03


It is a club that builds a human network and realizes self-management based on it.


Every other week, club invited lecturers on various topics such as dream, future, and self on Saturdays, and conducted self-development study with people including Java study and PT practice twice a week.


In this club, I have planned and conducted various programs such as famous restaurant reviews / affectionate villages / bucket list.


Curseware

2011.04 ~ 2016.08


Band club in Chung-Ang University


Performed as a singer and festival with other schools and acted as a vocalist.


Netory

2012.04 ~ 2016.08


As a society within Chung-Ang University, a club that mentors subjects to students


Mentoring on "linear algebra" courses and how to apply them to computer engineering

Work Experience

Data Engineer Assistant, Qualcomm, 2017.04 ~ 2017.06

Research for Data preprocessing

다운로드 (1).png

AI Research Intern, AI Research Institute, 2017.07 ~ 2017.11

Research for GAN, Image classification, Medical AI

logo.JPG

AI Research Intern, Lunit, 2017.11 ~ 2018.01

Research for Data Imbalance, Medical AI

logo.JPG

AI Research Intern, Naver Webtoon, 2018.01 ~ 2018.05

Research for Unsupervised Image to Image Translation GAN

logo.JPG

AI Research Engineer, NCSOFT, 2018.05 ~ 

Research for Vision AI, Working as a substitute of mandatory military service

logo.JPG

Research Experience

KISTI, 2015.04 ~ 2015.12

Research for natural language processing project about national R&D data

키스티.jpg

Korea Research Foundation, 2015.05 ~ 2017.04

Research for Social Media Curation Platform

한국연구재단.jpg

Pulson Information Technology & Institute of Law Intelligence Information, 2016.04 ~ 2016.08

Research for legal advisory robo advisor project

법률바다.png

Etc

Korea IT Business Promotion Association, 2013.12 ~ 2014.02

ICT Industry-University Cooperation Curriculum Hanwha S&C 

Completion of DB / Network course

Experiences 00 00@2x 504900dc09d82f711fdb54cf5763251cab5193a844856da978b2b8f3907ecd5a

Naver, 2017.05.26 ~ 2017.05.27

Naver Campus Hack Day - Perform for Keyword-analysis Project 

Selected as Excellent Participant

Experiences 00 01@2x 6daa47d1df70d315f129dcecdd562003831581c4cdd8f9c819aa22d4a24ae150

Presentation

Naver Tech-talk, 2018.01.08

Medical AI Tutorial via LUNA16 :: [video] :: [slide]

Experiences 00 00@2x 504900dc09d82f711fdb54cf5763251cab5193a844856da978b2b8f3907ecd5a

NVIDIA AI Conference, 2018.11.07

Image-to-Image Translation :: [link] :: [slide]

Experiences 00 00@2x 504900dc09d82f711fdb54cf5763251cab5193a844856da978b2b8f3907ecd5a

Implementation (2017)

Vector Similarity :: Java & Python :: [code]

Word2Vec :: Java :: [code]

TF-IDF:: Java :: [code]

Keyword Analysis :: Java :: [code]

Deep Convolutional Generative Adversarial Networks :: Tensorflow :: [code]

Learning to Discover Cross-Domain Relations with GAN :: Tensorflow :: [code] :: [slide]

Unpaired Image-to-Image Translation using Cycle-Consistent GAN :: Tensorflow :: [code]

Triple Generative Adversarial Nets :: Tensorflow :: [code]

Densely Connected Convolutional Networks :: Tensorflow :: [code]

Aggregated Residual Transformations for Deep Neural Networks :: Tensorflow :: [code]

Squeeze-and-Excitation Networks :: Tensorflow :: [code]

Network In Network :: Tensorflow :: [code]

Implementation (2018)

Curriculum Adaptive Sampling for Extreme Data Imbalance :: Tensorflow :: [code] :: [presentation]

On the convergence of Adam and Beyond :: Tensorflow :: [code]

Unsupervised Image-to-Image Translation Networks  :: Tensorflow :: [code]

Stabilizing Adversarial Nets with Prediction Methods  :: Tensorflow :: [code]

Tensorflow DatasetAPI Tutorial  :: Tensorflow :: [code]

Image-to-Image Translation with Conditional Adversarial Networks  :: Tensorflow :: [code]

Group Normalization  :: Tensorflow :: [code]

Spectral Normalization for Generative Adversarial Networks  :: Tensorflow :: [code]

Multimodal Unsupervised Image-to-Image Translation  :: Tensorflow :: [code]

Self-Attention Generative Adversarial Networks  :: Tensorflow :: [code]

StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation  :: Tensorflow :: [code]

Differentiable Learning-to-Normalize via Switchable Normalization  :: Tensorflow :: [code]

The relativistic discriminator: a key element missing from standard GAN  :: Tensorflow :: [code]

CartoonGAN: Generative Adversarial Networks for Photo Cartoonization  :: Tensorflow :: [code]

Generating a Fusion Image: One’s Identity and Another’s Shape  :: Tensorflow :: [code]

Identity Mappings in Deep Residual Networks  :: Tensorflow :: [code]

Diverse Image-to-Image Translation via Disentangled Representations  :: Tensorflow :: [code]

Batch-Instance Normalization for Adaptively Style-Invariant Neural Networks  :: Tensorflow :: [code]

Partial Convolution based Padding  :: Tensorflow :: [code]

Implementation (2019)

Large Scale GAN Training for High Fidelity Natural Image Synthesis :: Tensorflow :: [code]

Tensorflow Cookbook :: Tensorflow :: [code]

Adaptive Gradient Methods with Dynamic Bound of Learning Rate :: Tensorflow :: [code]

Current Interests

Quotes 02 00@2x ec97a834b559e4f4024fd1b628ddb62c287bedbc8084d8c8c7c08d59b9b930ee

Machine Learning & Deep Learning


Unsupervised Image to Image Translation


Generative Adversarial Network