Hi, I’m Daniel Baek.
Welcome to my personal site
Scroll down to see my works
[01]
08-2025
Graphical Neural Networks
[Graphs, Rail, Delay Propogation]
This review finds that since 2020, graph neural networks using network structure, temporal patterns, and railway data enable minute-scale rail disruption prediction and re-routing, demonstrated in UK, Dutch, and French deployments despite data and benchmarking challenges.
[02]
12-2023
Head-Mounted IMU
[Programming, Circuitry]
Using a circuit-board-mounted IMU and two potentiometers to filter the signal, I developed a head-mounted mouse able to control inputs on a computer screen.
[03]
06-2025
Object Detection for Metro Rail
[Computer Vision, Machine Learning]
This paper introduces a YOLOv8n-based AI that detects mobility aids to automate transit doors and ramps, achieving 0.9946 precision, 0.9846 recall, and 0.9893 F1—outperforming VGG16 for real-world deployment.
[04]
05-2024
Bikeshare Efficacy
[Statistics]
Using 2019–2023 data, I found LA’s bike-share growing but unevenly covered and underfunded, concluding it needs more expansion, integration, and user-focused upgrades to be Olympics-ready.
[05]
12-2023
Object Detection for Cars
[Object Detection, Machine Learning]
I used AI object detection (sliding windows, CNNs, VGG16, YOLO), showing accuracy gains up to ~95% via transfer learning and real-world uses like autonomous vehicles