Machine learning intro
Undergraduate · CS / Programming
Syllabus focus
Topics typically covered
Standard syllabus
Foundations
- Learning problems: classification, regression, clustering
- Train/validation/test splits and cross-validation
- Bias–variance tradeoff and model selection
- Linear and logistic regression
- k-nearest neighbors and naive Bayes
Core methods
- Decision trees and ensemble methods (random forests, boosting intro)
- Support vector machines (intro)
- Neural networks: perceptron, MLP, backprop (intro)
- Clustering: k-means, hierarchical (intro)
- Dimensionality reduction: PCA (intro)
Evaluation and responsibility
- Train/validation/test splits and leakage pitfalls
- Precision, recall, F1, ROC/AUC tradeoffs
- Bias/variance intuition and regularization
- Feature engineering vs representation learning (survey)
- Ethics: fairness, privacy, and dual-use awareness
- Reproducibility: seeds, versions, and data cards (intro)
STEM / applied
Implementation
- Scikit-learn pipelines: preprocessing, fit, predict
- Feature engineering and handling categorical variables
- Hyperparameter tuning with grid/random search
- Model evaluation metrics: accuracy, F1, ROC-AUC, RMSE
- Visualization of decision boundaries and embeddings
Applied ML
- Working with imbalanced data and leakage pitfalls
- Intro to deep learning frameworks (PyTorch/TensorFlow survey)
- Responsible AI: fairness and interpretability (intro)
- Deploying models as batch or API services (intro)
- Case studies: text, vision, or tabular domains
Pipelines and deployment intro
- scikit-learn style fit/predict workflows
- Cross-validation and hyperparameter search (intro)
- Handling imbalanced classes
- Model serialization and simple inference services
- Monitoring drift after deployment (survey)
- Capstone: end-to-end supervised learning mini-project
Notes
Prerequisites usually include linear algebra, probability, and programming. Math depth varies widely.