MLEARN520-Notes
Course 2: Advanced Machine Learning
These are my personal study notes for the second course in the Machine Learning Professional Certificate Program from the University of Washington. The lectures are based on the book Introduction to Statistical Learning by James, Witten, Hastie, and Tibshirani, and Data Mining, Fourth Edition: Practical Machine Learning Tools and Techniques by Witten, Frank, Hall, and Pal.
Table of Contents
| Week | Topic | Concepts |
|---|---|---|
| 1 | Course Intro and Review | Multi-class Classification |
| 2 | Decision Trees | Classification Trees, Regression Trees |
| 3 | Tree Emsemble Methods | Bagging, Random Forests, Boosting, Bayesian Additive Regression Trees |
| 4 | Support Vector Machine (SVM) | Support Vector Classifier, Support Vector Machines, Support Vector Regression |
| 5 | Stacking and Blending | Stacking, Blending |
| 6 | Clustering Methods | K-Means Clustering, Hierarchical Clustering, Gaussian Mixture Models |
| 7 | Natural Language Processing (NLP) | Vector Space Models, Probabilistic Models, Sequence Models, Attention Models |
| 8 | Recommendation Systems | Collaborative Filtering, Content-Based Filtering, Hybrid Filtering, Knowledge-Based Filtering |
| 9 | Model Interpretation | Types of Interpretability, Common Interpretability Techniques |
| 10 | MLOps | MLOps Workflow, Model Versioning and Data Management, Continuous Integration and Continuous Delivery (CI/CD), Model Monitoring |
Homework Assignments
Unlike the lectures, which takes a theoretical approach to machine learning, the homework assignments are more hands-on and practical. The assignments are done in Python using Jupyter Notebooks with extensive of the scikit-learn library.
For academic integrity reasons, I won't be posting my code here. However, I will provide a brief overview of the key concepts covered in each homework assignment.
Clarification on Content Creation
While my understanding of the subject matter originates from attending the class, the majority of my notes were based on the provided slides. These notes can be a bit messy and may potentially breach copyright terms. To have a digital, easily accessible version of my notes, I opted to work with ChatGPT to refine and present the content in a more organized manner. Given that these topics are common knowledge and widely discussed online, the information rendered by ChatGPT is generally trustworthy.