Purdue University: Introduction to Scientific Machine Learning Course Syllabus

Overview: This advanced course introduces learners to Scientific Machine Learning (SciML), a cutting-edge field combining machine learning with scientific computing. The curriculum emphasizes real-world applications in engineering and physics, focusing on methods like physics-informed neural networks (PINNs). Designed for those with strong mathematical and programming backgrounds, the course spans approximately 15–20 hours of learning, including hands-on labs, case studies, and assessments. Learners will gain practical skills in building and evaluating models using scientific datasets, preparing them for research or advanced roles in computational science and AI.

Module 1: Data Exploration & Preprocessing

Estimated time: 3 hours

• Introduction to key concepts in data exploration & preprocessing
• Hands-on exercises applying data exploration techniques
• Review of tools and frameworks commonly used in practice
• Best practices for handling scientific datasets

Module 2: Statistical Analysis & Probability

Estimated time: 3–4 hours

• Foundations of probability and statistical inference
• Case study analysis with real-world scientific examples
• Application of statistical methods to extract insights
• Discussion of industry standards and best practices

Module 3: Machine Learning Fundamentals

Estimated time: 4 hours

• Core concepts of supervised and unsupervised learning
• Interactive lab: Building practical ML solutions
• Case study analysis with real-world applications
• Implementation using industry-standard tools

Module 4: Model Evaluation & Optimization

Estimated time: 2–3 hours

• Techniques for evaluating model performance
• Methods for hyperparameter tuning and optimization
• Guided project work with instructor feedback
• Review of frameworks used in scientific ML workflows

Module 5: Data Visualization & Storytelling

Estimated time: 2 hours

• Principles of effective data visualization
• Interactive lab: Communicating insights through visuals
• Case study analysis with real-world examples
• Guided project work integrating storytelling into analysis

Module 6: Advanced Analytics & Feature Engineering

Estimated time: 1–2 hours

• Feature engineering for scientific datasets
• Advanced analytics techniques in ML pipelines
• Interactive lab: Building practical solutions
• Review of tools and frameworks for large-scale data

Prerequisites

• Strong background in mathematics (linear algebra, calculus, differential equations)
• Proficiency in Python programming and numerical computing
• Familiarity with machine learning frameworks and scientific computing concepts

What You'll Be Able to Do After

• Build and evaluate machine learning models using real-world scientific datasets
• Design end-to-end data science pipelines applicable to engineering and physics problems
• Apply statistical and ML methods to extract insights from complex, multidimensional data
• Utilize physics-informed neural networks (PINNs) for scientific simulations
• Communicate analytical findings effectively through visualization and storytelling