Taipei, Taiwan
My name is Ken Lin and my Chinese name is 林楷宸. I'm a graduate student at National Taiwan University. My current research interests are Federated Learning, Machine Learning Security and Adversarial Attacks. I'm also interested in Artificial Intelligence of Things and Reinforcement Learning. Our project, Vision-Based Autonomous Navigation Robot, received second place in the world in NVIDIA AI at the Edge Challenge.
Sep 2020 - present.
GPA: 4.19 / 4.3
Sep 2016 - Jan 2020
GPA: 4.17 / 4.3 (Rank: 4 / 134)
Academic Awards - Spring 2018, Fall 2018
Jul. 2018 - present
Taipei, Taiwan
Adviser: Dr. Shang-Tse Chen (AI Security Lab)
Research on machine learning security, especially on the transferability of adversarial attack. Currently, my research topics also include the security of federated learning.
Jul. 2020 - Sep 2020
Taipei, Taiwan
Develop a web-based tool interface using React.js and Redux.js. Utilize Tornado, a python web framework library, to build an asynchronous storage server, which is integrated with MongoDB.
Jan. 2018 - Jul. 2020
Hsinchu, Taiwan
Adviser: Dr. Chun-Yi Lee (Elsa Lab)
Research on reinforcement learning and simultaneous localization and mapping (SLAM). Build an unmanned aerial vehicle with only a monocular camera using computer vision and reinforcement learning. Develop a learning-based visual odometry method, for estimating the ego-motion of a monocular camera.
Jul. 2017 - Aug. 2017
Hsinchu, Taiwan
Adviser: Dr. Hwann-Tzong Chen
Edit C and C++ course materials and sample codes for freshmen in CS, NTHU.
Feb. 2018 - Jul. 2018
Aug 2018 - Jan 2019
Aug 2019 - Jan 2020
Feb 2019 - Jun 2019
Feb 2020 - Jun 2020
Design an app to prevent people from impulsive consumption. The app is developed using Javascript, React Native and Redux.js. We also deploy a Node.js backend server on Amazon Web Services using Docker.
Build an effective and low-cost modular framework for navigating a robot via virtual signals. Introduce "Virtual Guidance" and use reinforcement learning for robot navigation tasks. This project received 2nd Place in the world in NVIDIA AI at the Edge Challenge. [link]
Build a line bot to record and report the current number of users in NTU gym, preventing users from going to the gym full of people. Develop the server using Django, Docker-compose and Line chatbot API. The final goal of this project is to predict the daily usage of the gym.