TUTOR

M07 · Level II — Machine Intelligence

Classical Machine Learning

Understand classical machine learning as statistical model construction, estimation and generalisation—not library operation.

Topics

  • Statistical Learning
  • Regression
  • Classification
  • Trees
  • Ensembles
  • Margins
  • Clustering
  • Dimensionality Reduction

Technical concepts

  • 01What Makes a System a Learning System?
  • 02Supervised Learning Formulation
  • 03Linear Regression from First Principles
  • 04Residual Analysis & Regression Assumptions
  • 05Regularised Linear Models
  • 06Bias, Variance & Model Complexity
  • 07Logistic Regression
  • 08Classification Boundaries & Thresholds
  • 09k-Nearest Neighbours
  • 10Decision Trees & Recursive Partitioning
  • 11Gini Impurity & Entropy
  • 12Tree Complexity, Pruning & Generalisation
  • 13Ensemble Learning: Bagging
  • 14Boosting
  • 15Support Vector Machines & Maximum Margin
  • 16Kernel Methods
  • 17Naive Bayes
  • 18K-Means from First Principles
  • 19Clustering Validation & Stability
  • 20Hierarchical Clustering
  • 21PCA: Variance-Preserving Representation
  • 22Feature Selection vs Feature Extraction
  • 23Generative vs Discriminative Classical Models
  • 24Baselines & Model Comparison
  • 25Model Error Decomposition
  • 26Research & Industry Transfer

Practical implementation

  • 01Laboratory 01 - Classical ML Baseline Stack
  • 02Laboratory 02 - Clustering & Representation
  • 03Laboratory 03 - Model Stability & Perturbation
  • 04Laboratory 04 - Explainable Baseline Report

Tools and technologies

  • reproducible VM/container
  • Python + NumPy/ SciPy + scikit-learn
  • experiments and diagnostics

Project

Laboratory 04 - Explainable Baseline Report

Expected learning outcomes

  • 01Formulate supervised and unsupervised learning problems.
  • 02Derive linear regression and logistic regression objectives.
  • 03Explain regularisation and model complexity.
  • 04Understand decision-tree induction.
  • 05Understand ensemble learning.
  • 06Understand maximum-margin methods and kernels.
  • 07Perform clustering as structure discovery.
  • 08Use PCA for representation.
  • 09Build an evidence-based ML baseline.

Assessment

  • 01Daily practice — 20%
  • 02Laboratories — 30%
  • 03Research reproduction — 10%
  • 04Integrated project — 20%
  • 05Technical viva — 15%
  • 06Engineering record — 5%

Source document: TUTOR Frontier Module 07 Classical Machine Learning