Testimonial Essay _  Yi Lu from Tokyo Institute of Technology

I have always been interested in Korea and the educational environment of the world's top-ranked KAIST. Not only for the educational environment but also as a research intern at the National Institute of Advanced Industrial Science and Technology, it was also interesting to compare the research environment of the two countries national research institutes.

I have taken part in the IBS data science project, which is also the lab at KAIST. Good research in a wide range of areas in various research fields, such as computer vision, data analysis of covid 19, medical images, and fake news. And data science lab actively engages in exchanges with research labs abroad and tries to explore the intersection of data science and different fields

 It was very good for me to experience the life of the school and the life of the institute. Even though my research area is not data science, I tried to work in this research lab.  I just take some basic AI and data science lectures before I came to Korea. This is a very good chance for me to practice my learning using a specific project. And AI security is a very popular area of research. In data science, we use a lot of data, which can contain a lot of personal information. Being able to understand more about the data science process will help me to read and study AI security better in the future.

Regarding my project, I started by reviewing my knowledge of basic neural networks to learning about convolutional neural networks. I also read three papers on super-resolution and learned about this scientific field. In late January, I started reading the code of my mentor's paper and trying to configure the development environment and run it. Formally, I moved on to the actual execution of the code and tried to change the dataset to test the existing model. My work schedule is like the following：

 12/19-12/30:

 Watching the video of the neural network

 learning the online course of  CS231n: Convolutional Neural Networks for Visual Recognition

 The famous construction of CV

1/3-1/5: Take a trip to Seoul

1/9-1/13: Reading three project-related papers

Image Super-Resolution Using Deep Convolutional Networks, ECCV 2014 (First Deep learning based Super Resolution)

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network, 2017 CVPR (First GAN-based Super Resolution)

Downscaling Earth System Models with Deep Learning

1/16-1/18: set up a development environment and run the code

1/19-1/22: Read the code and understand the construction of the Model and apply the model to another dataset

I am very grateful to this project for giving me the opportunity to really get involved in a data science project, even though I couldn't do more than that in the short time I had, it was a way for me to learn more. I learned about neural networks, convolutional networks, and super-resolution, and I learned about various CNN models that have good performance and are now widely used.

What I like most about the program is the high degree of freedom in the choice of research laboratories. It is very difficult for Ph.D. students to go to the opposite laboratory without having a suitable research direction with the opposite teacher. But this program makes it relatively easy to choose a research lab, which makes interdisciplinary communication very easy. And the people you come into contact within your life and in the research, lab are very international, most of them have overseas experience or are foreigners whose are also very broad minded. The professor's proactive approach to external communication has allowed me to meet different types of researchers and collaborate with researchers from different fields, which I also found very worthwhile to learn from.

Most companies in the security industry are now focusing their experience on AI security and being able to combine the AI fundamentals I have learned this month with the knowledge I have in the security field will greatly improve my competitiveness in my future career and research.

Most companies in the security industry are now focusing their experience on AI security and being able to combine the AI fundamentals I have learned this month with the knowledge I have in the security field will greatly improve my competitiveness in my future career and research.