CS2007 · Diploma / Data Science · 4 credits

Machine Learning Techniques

Supervised and unsupervised learning — regression, classification, trees, ensembles, clustering, dimensionality reduction.

Instructor Prof. Arun Rajkumar Term Every term Prereq CS2004

§ 1Calendar

  • Quiz 1—
  • Quiz 2—
  • End-term examination—

§ 2Previous Year Questions

  • MLT weekwise PYQs

    Past questions sorted by week.

    discourse.onlinedegree.iitm.ac.inby Preethy ParthasarathyDiscourse login required

§ 3Resources

Course material2

Notes7

Practice2

TA resources1

  • MLT TA sessions, T2 2026

    Recordings and materials from the TA sessions.

    discourse.onlinedegree.iitm.ac.inDiscourse login required

NumPy practice4

CS2007 · Diploma / Data Science · 4 credits

Syllabus

The official course syllabus, transcribed from the IITM BS Data Science & Applications course booklet.

§ 1Course Description

This course introduces the primary methods and models used in regression, classification, and clustering, and examines their properties to understand their suitability for different problem settings.

Lectures-Tutorials-Practicals-Credits: 4-0-0-4

§ 2Faculty

  • Prof. Arun Rajkumar, Wadhwani School of Data Science, IIT Madras

§ 3Week-by-week

Week 1: Machine Learning - learning paradigms; Representation learning; PCA.

Week 2: Unsupervised Learning - PCA limitations and complexity; Feature transformation; Kernel methods.

Week 3: Clustering - K-means algorithm; Convergence and cluster properties; Initialization methods (K-means++); Choosing number of clusters.

Week 4: Estimation - MLE, Bayesian; GMM and likelihood; Convexity and Jensen's inequality; EM algorithm.

Week 5: Supervised Learning - linear regression; Optimization (gradient descent); Kernel and probabilistic regression.

Week 6: Linear regression - evaluation, cross-validation; Bayesian modeling; Regularisation (Ridge, Lasso).

Week 7: Classification - K-nearest neighbours, decision trees; Generative vs discriminative models.

Week 8: Generative models - Naive Bayes (Gaussian, decision function); Limitations.

Week 9: Discriminative Models - perceptron (update rule and convergence); Sigmoid function; Logistic regression.

Week 10: Maximum margin methods - SVM, soft margin; Dual formulation; Constrained optimization.

Week 11: Soft-margin SVM - dual formulation and complementary slackness; Overfitting and underfitting; Ensemble methods - bagging and boosting.

Week 12: Loss functions - SVM, logistic, perceptron, boosting; Neural networks; Summary.

CS2007 · Diploma / Data Science · 4 credits

Weekly notes

Week-wise lecture notes and activity question solutions for Weeks 1 to 12, shared by Piush Das.

§ 1Notes and solutions

Lecture notes12

Activity question solutions12

Question compilations2

CS2007 · Diploma / Data Science · 4 credits

Modified slides

Lecture slides reworked by Piush Das for clarity. They reflect his own understanding and may contain mistakes, so use them alongside the official slides, not instead of them.

§ 1Week-wise slides

Modified lecture slides12

Piush Das is still updating these as the term goes on, so later weeks may change. If you find an error, tell him on the Discourse post.