Personal Projects
In addition to my research, you can also find some of my personal projects on this website. Their repositories can be found on my Github page!
Latest Updates
Leaftracker is an image classification model that differentiates toxic plants from their non-toxic look-alikes. Links to my huggingface.co page, Github repository, and info pages are linked below!
Formally Verified PID Controller
Applying CBMC to perform bounds checking for a simulated PID controller in C
I'm currently building a modular PID control loop in C and formally verifying key safety properties using CBMC, such as proving bounded outputs, robustness to variable sensor inputs, and stability across time. I've been automating SAT-based verification checks using GitHub Actions and plan to continue the project by supporting future STM32H753 integration, such as verifying certain pin bounds checking.
SmartQuake
A machine-learning model for long-term earthquake prediction
- 4000 Years
- >200k Events
- 25 Observatories
Strong earthquakes are infrequent yet destructive events, mainly due to their unpredictability. Earthquake warning systems can only reactively send an alarm upon detection of an earthquake, which leaves only a small time frame for people to head to safety. SmartQuake is a project that performs long-term earthquake prediction using a time series model, inspired by the TimesNet architecture.
The kepler.gl map plots significant earthquakes recorded by the Canadian National Earthquake Database, where magnitudes are >= 4.0. Many databases have been scraped and compiled to construct the neural network model.
About Me
Hey there! My name is Luke Nam, an incoming computer science Ph.D. student at Vanderbilt University. My main research interest lies in formal verification of neural networks, and verifying autonomous cyber-physical systems (CPSs). You can find my research lab contributions throughout this website! Outside of academics, I enjoy watching college basketball and was a former clarinet in the Duke University Marching Band.
