Sunner Li
"I do not just study for study but being enthusiastic for the dream! In the future, I will keep using my enthusiasm and ability to solve some problems, do some cool things, and even change this world for a little bit." Starting to get touch with deep learning from college and explore the generative model in the graduate school. My experience includes the application of computer vision and natural language processing with deep learning.
Student in NCTU
Taichung，TW
[email protected]
Education
National Chiao Tung University
Master of the institude of computer science and engineering
(2017/09 - )
National Taipei University
Bachelor of the computer science and information engineering
(2013/09 - 2017/08)
Skill
Deep learning
- Tensorflow
- PyTorch
Language
- C & C++
- Python
- English (TOEIC 845)
Publication
[1] Bor-Shing Lin, Cheng-Che Lee, and Pei-Ying Chiang, "Simple Smartphone-Based Guiding System for Visually Impaired People", Sensors, vol. 17, no. 6, pp. 1371, 2017.
Open Course
The Introduction of Natural Language Processing
University of Michigan
(2016/09 - 2017/01)
The Machine Learning Foundation
National Taiwan University - prof. Hsuan-Tien Lin
(2016/09 - 2016/11)
Prize
- 國立台北大學書卷獎七次
- 國立台北大學飛鳶獎
- 音律電子股份有限公司獎學金七次
- 國立台北大學電機工程學系專題展第一名及人氣獎
- 國立台北大學資訊工程學系專題展第二名
- 入選科技部大專生研究計畫
Project Experience
( The list of Group Project )
Gerli - An integration accounting system based on sentiment analysis
(2016/09 - 2017/01)
This is the final project of the software engineering course which is made with several young students. This program can do the semantic analysis toward the input accounting string by the remote computer, and send the appropriate response string to the mobile phone. This design can make the user to change the spending habit.
Cup2
(2017/06)
This is the final program of computer vision. The program use CNN which is inspired by YOLO to do the pattern segmentation and pattern recognition. At last, the boundary and the name of pattern class would be shown in the screen.
FastChi
(2017/12)
This is the final program of the parallel program. In this project, we try to use 2 GPU to accelerate, and adopt the idea of inception to enhance the original render network. By this changing, it speed up for 1.7 times than the origin. At last, we purposed the idea to use YOLO network to do the style extraction network.
( The list of Personal Repository)
MySimpleGirlfriend_2017
(2017/05 - 2017/06)
I build a simple girlfriend which can do the simple conversation on facebook. The program use facebook chatbot API and seq2seq concept which is implemented by tensorflow API. By the training of my collected data, the simple conversation can be achieved.
Simple-Hourglass
(2017/10)
This project use tensorflow to implement and discuss the concept of pixel-by-pixel segmentation. In this work, there are five different model which had been discussed, and I implement the three model, including FCN-8, U-Net and RedNet. Toward the data collected by myself, the RedNet structure can recognize the object successfully.
Waiting-for-you generator
(2018/01)
This project use CycleGAN structure to do the style transfer toward the image space in the new song of Jay Chou - waiting for you. Additionally, I also try the structure with another mask (Mask-CycleGAN) to remove the influence of the lyric during training. The result shows that it can learn the dark and pastel-like style.
Bio
My name is 李政哲 (Cheng-Che Lee), and the English name is sunner. From the 2nd year of college, I spend time coding for interesting mobile application, and meet with deep learning in 3rd year. In my graduate project, we integrate the Faster-RCNN which was written in caffe and become the simple guiding system that can be used by mobile phone. Additionally, I write some related application with deep learning framework in my 4th year. As the result, I have the application experience of tensorflow and pytorch.
The deep learning is an advance territory. To more realize the principle, I participate the introduction of artificial neural network in NTPU and listen the material of the machine learning foundation which was lectured by prof. Hsuan-Tien Lin in my last year of college. Moreover, the machine learning course which was taught by NCTU is also majored by myself. Through keeping learning, I have the fundamental idea toward deep learning.
With the experience of my second-year project, I know some Linux command and it makes me smoothly install the environment of deep learning. Additionally, several times of group working experience equip me with independent planning ability toward the project. At last, I also have some experience about git.
It’s very important to have self-learning ability. During my 4th year of college, I also major the introduction of natural language processing course which was lectured on Coursera platform and get touch with some idea of NLP. After going to graduate school, I read some deep learning paper by myself, including object detection, object segmentation, mobile model, generative model and style transfer. In addition to simplified implement the concept of several paper, I finally make an interesting style transfer generator in the winter vacation.
After completing several works, I realize that I can use my ability to change this world. As the result, I hope that I can keep working on this area. In this fast-changing era, there’re also a lot of model and concept which I didn’t know. I wish that I can have more clear understanding in deep learning territory, learn with latest knowledge and make much coolest work. After I join the team, I wish that I can use my ability to contribute for yourself!
