Welcome to my website!

About me

My name is Aukkawut Ammartayakun (/ˈɔː.kə.wuːt/ /ɑːm ˈmɑ:d ˈtɑː jə ˈkʊn/). I am a Ph.D student at the University of Tennessee, Knoxville. I study data science and love learning new things about DS. My current research interest is in the robustness of intelligence systems. This includes but is not limited to, adversarial attacks and multi-agent systems. I am also interested in the medical and biological applications of mathematics and statistics.

I essentially interested in the theory of how to make AI system powerful while still figuring out how to shut it down. I am also interested in applied field where I use math to actually do something that is not math. I like to torture people (and AI) in other fields with Lipschitz condition basically.

If you want to access my CV, please click here.

Education

• High School
  • Science Classroom in University - Affiliated School (SCiUS) program at Silpakorn University (2016 - 2019)
  • Postgraduate at Northfield Mount Hermon (2019 - 2020)
• College
  • B.S Data Science at Worcester Polytechnic Institute (2020 - 2024) [With distinction]
  • M.S Data Science at Worcester Polytechnic Institute (2023 - 2024)
  • Ph.D. Data Science and Engineering at The University of Tennessee, Knoxville (2024 - Present)

Research and Publications

If the pre-print is unavailable, please email me to request one, and I will gladly send it to you.

• Conferences
  • [Best Paper Award] Aswale, A., Lopez, A., Ammartayakun, A. and Pinciroli, C., 2022. Hacking the Colony: On the Disruptive Effect of Misleading Pheromone and How to Defend Against It. In: 21st International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2022). IFAAMAS. Conf PDF
  • [Best Paper Award] Krastev, S., Ammartayakun, A., Mishra, K. J., Koduri, H., Schuman, E., Morris, D., Feng, Y., Bandi, S. S. R., Ngan, C.-K., Yeung, A., Li, J., Ko, N., Emdad, F., Rundensteiner, E., Ho, H. M. H., Wong, T. K., Chan, J. P. C., 2024. META: Deep Learning Pipeline for Detecting Anomalies on Multimodal Vibration Sewage Treatment Plant Data. In: 16th International Conference on Neural Computation Theory and Applications (NCTA 2024)
• Preprint
  • Busaranuvong, P., Ammartayakun, A., Korkin, D., and Khosravi-Far, R., 2023. Graph Convolutional Network for Predicting Secondary Structure of RNA preprint

Work (related) Experiences

• 2008 - 2012
  • Helping in a live educational TV broadcast channel, including transitioning between tapes, controlling sounds, playing/recording/handling tapes, etc.
• 2019 - 2020
  • Making the bakery (and prepare multiple of 1 pound 6 ounces pizza dough) for students in the high school.
  • Cleaning trophy and hallway in the high school.
  • Preparing the banquet for athletes in the school.
• 2022
  • Research Assistant in EREE program at WPI.
    • Working on simulating the topoisomerase mechanism in elastic rod and fluid coupled system.
  • Teaching Assistant in CS 525/DS 595 Reinforcement Learning in Fall 2022 at WPI.
    • Helping with technical issues, grading quizzes and assignments, hosting office hours, and answering RL-related questions.
• 2024
  • Research Assistant in Bredesen center, UTK.
    • Working in Data Science and Engineering (DSE) program

Classwork

You can request access to my transcript here (please write the reason on the file access request form in Google Drive): transcript. Notable classes are listed here (by their description rather than names).

• CS/Engineering : Swarm Intelligence, Natural Language Processing, Reinforcement Learning, Machine Learning, Database Management Systems
• Math/Stats: Statistical Inferences and Probability, Statistical Learning, Optimization for Deep Learning, Causal Inference, Multivariate Analysis, Bayesian Statistics, Stochastic Processes
• Biomedical related: Computational Neuroscience, Clinical Psychology and Mental Health
• Business: Prescriptive Analytics, Machine Learning in Business, Operation Research

Note: I intentionally reduce my GPA on my Master degree. I can choose the class (among class I already taken) that I get A, but I don’t. Just because I don’t think I know enough to have near-perfect GPA.

Learning Philosophy

Recently, I have heard the saying “let them cook”. I think I am more align with this saying. I learn by trying not by telling. Think about it for a second, we probably know that stochastic gradient descent (SGD) and its derivative like ADAM are crucial method for optimizing neural network. However, why is that? Because textbook and professor told you so? That does not satisfy my curiosity. So, to fully understand the method of optimizing neural network, I will try different method like using Bayesian optimization. Maybe even try particle swarm optimization to see why exactly people use ADAM more than something that, in theory looks better. I hope I can incorporate this saying to teach students too.

But obviously, this method does not work for… example, knowing why we use Uranium for atomic bomb. Like, you don’t have to experiment exploding Plutonium nuclear bomb to see it. This philosophy is more towards exploration both experimentally (hands-on) and theoretically (i.e., open your mind and see if some assumption is dropped, what happens). Simulation exists and even if there is none, you can make one.

If I were to put this for teaching students, I would let them explore the project both supervised and unsupervised. For example, I might have a planned project where I pick the topic and oversee students’ trials and error. Like, letting students directly collecting satellite image data using the dipole antenna to reconstruct the full satellite image and use that for the analysis like weather forecasting. I might supervise the image collection but I might let them try to find the way to predict the weather by themselves. Maybe use CNN and make your own dataset from weather forecasting API? Maybe use image processing and statistical method to summarize that image down to numerical value and have a rule-based classifier to predict the weather.

Presentations and Projects

Presentation

Most of the presentations/lectures I made can be found here:

• SCiUS @ Silpakorn University (High school level, lectured with the Thai language)
  • SC 30253 Biotechnology: Plant nutrients for tissue culture link
  • SC 30253 Biotechnology: Microorganisms and the treatment for cancer link
  • Physics 4: Nuclear Power Plants link
• WPI
  • Seminar talk on Few-shot learning link
  • My part of the lecture on Neural encoding link

Projects

Here is the list of some class/graduation projects I have done (please look into my github for more recent projects):

• Graduation Projects
  • Exploring RNA Secondary Structure Prediction using Graph Convolutional Neural Networks (Graduation project)
  • Westborough High School Mental Health Predictive Analysis (Graduation project)
  • Transformer-based Anomaly Detection for Multiple Sources of Vibration Data (Graduation project)
    • Video link
• Class Projects
  • Using generative adversarial networks for recovering the information from inverse Conway’s game of life
  • Fine-tuning transformer model with adversarial training for news headline generation
    • Proposal link
  • Adversarial news propagation and stochastic programming simplification
    • Presentation link
    • Website link
  • I hate machine learning classification, so I use hypothesis testing to classify MNIST data.
    • Website link
  • Silly Dimensional Reduction Algorithm
    • Website link
• Other Projects
  • Finding the mechanical force required by Topoisomerase II to change DNA topology under the fluid

Contact

Contact Information

Write “Urgent” on your email title if you need to have my immediate response. I will try my best to answer emails in a timely manner.

Email: aammartayakun [at] tennessee.edu

Personal email (like gmail) and social networks (like Facebook) will be reserved to only closed friends and colleagues.

Please note that the message sent to my personal account like Line or Messenger with regard to work will be rejected.