Pioneer Academy

Outreach
High School Research
Data Science
Two months of summer school for six students, teaching Deep Learning and AI, and six individual research projects.
Published

June 1, 2025

Participants:

  • Tairan Ma: “Multimodal Machine Learning for Physiological Privacy Protection in Facial Videos”
  • Yichen Leng: “Data Analytics for Speed-Cubing Rankings”
  • Mohammed Ali: “Machine Learning for Memristor Networks”
  • Xiaodan (Amy) He: “Convolutional Neural Networks for Facial Expression Recognition: Analysis of Current CNNs and Integration of AI Enhancement”
  • Terry Duan: “Retrieval-Augmented Generation for Intelligent Learning Tools”

Syllabus

Program Title: Machine Learning applied to Scientific Problems Pioneer Syllabus Term: Summer 2025

Program Topic Description:

This course introduces the students to Machine Learning methods to solve scientific problems. Traditionally, every scientific field has evolved its own methods to tackle problems – whether it is modeling and solving differential equations, using statistics, doing regression analysis and figure out prominent trends etc. Examples of these methods are found everywhere from physics, chemistry, biology, mathematics, astronomy, and even newer fields like climate analysis, renewable energy systems. With the advent of data driven methods and Machine Learning, many of these same methods are merged and most scientific problems are now attacked by employing all these methods in combination.

We will be using a combination of lecture notes uploaded on the LMS, pen-paper calculations, and running Python codes via Jupyter Notebooks.

We will then go through several standard machine learning methods:

  • Broad Types: Supervised and Unsupervised
  • Linear and Logistic Regression
  • Classification methods: Decision Trees, Random Forests, K-Nearest Neighbors, Support Vector Machines.
  • Artificial Neural Networks: Feed Forward network, Convolutional Neural Network, Recurrent Neural Network.
  • Physics Informed Neural Networks and Echo State Networks.
  • Time Series Methods: LSTM, GRU.
  • Large Language Models and Transformers.

These methods will allow us to tackle pretty much any problem in scientific fields, and they also form the basis of solving very serious problems like cancer detection to detecting livable exoplanets. In individual sessions, students will consult with the instructor and create a project that best aligns with their interests. It could be in any scientific field. The project should involve one of the machine learning methods applied to study the problem. The student will write up a report/paper about their findings. Preferred writing software is LaTeX for scientific writing, but Word is acceptable. If you want to use LaTeX, I can provide a tutorial on how to use it.