Machine Learning

Course Offering

Course Overview

This is an introductory course on Machine Learning (ML) that is offered to undergraduate and graduate students. The contents are designed to cover both theoretical and practical aspects of several well-established ML techniques. The assignments will contain theory and programming questions that help strengthen the theoretical foundations as well as learn how to engineer ML solutions to work on simulated and publicly available real datasets. The project(s) will require students to develop a complete Machine Learning solution requiring preprocessing, design of the classifier/regressor, training and validation, testing and evaluation with quantitative performance comparisons.

Course Objective

Explain the different types of learning problems along with some techniques to solve them.
Model real-world problems, apply different learning techniques and quantitatively evaluate the performance.
Identify and use advanced techniques through existing machine learning tools and libraries.
Analyze performance of ML techniques and comment on their limitations.

Course Syllabus

Introduction: Introduction to Machine Learning; Deep Learning vs Machine Learning; Review
Empirical Risk Minimization: Learning Problems and the Risk Minimization Framework
Regression: Linear and Logistic Regression
Machine Learning in Practice: Bias-Variance; Training/Testing; Overfitting; Jackknifing/Cross-Validation; Occam’s razor; Regularization and Model Selection
Classification: Naïve Bayes; Decision Trees; Random Forests
Neural Networks: Perceptron; Backpropagation; Training Neural Networks
Support Vector Machines: Optimization basics; Primal Form; Dual Form; Kernel SVMs, Soft-Margin SVMs
Unsupervised Learning: k-means clustering; Spectral Clustering (if time permits)
Deep Learning: Convolutional Neural Networks and applications to Images and Speech
Ethical Issues: Fairness, Accountability, and Transparency in Machine Learning

Textbooks

Machine Learning, Tom M. Mitchell, McGraw Hill, 1997
Pattern Classification, Duda, Hart and Stork. 2nd ed., Wiley, 2006
Pattern Recognition and Machine Learning, Christopher M. Bishop, Springer, 2006
Understanding Machine Learning: From Theory to Algorithms, Shai Shalev-Shwartz and Shai Ben-David, Cambridge University Press, 2014
Neural Networks and Learning Machines, 3rd Edition, 2009

Related Journals and Conferences [!Exhaustive]

Course Administration

Google Classroom will be used as primary mode of course administration, including general announcements, announcements/submission of assignments, doubt resolution.
Colab will be used for Coding Assignments.
Course will be supported with teaching assistants (TAs). TAs will communicate their office hours individually.
TAs will support the course through doubt resolution, responsible for announcement and/submission of quizzes/assignments, grading with faculties supervision etc.

Plagiarism Policy

The course follows a zero tolerance on academic dishonesty. The course follows the IIIT-Delhi policy on Academic Dishonesty
[Updated Institute policy will apply]
All plagiarism cases will be on record for your tenure at IIIT-Delhi
All code and reports will be checked for plagiarism
Home work theory questions may be asked in exam or quiz as it is
If correct in Home work and incorrect in exam, home work question will be marked zero. Make sure you know all your home work theory questions.

Auditing Requirements

Students need to submit all assignments and projects on time.
Students should achieve a grade of at least the class average.
Audit students are not permitted to do projects with registered students.
Quizzes and Exams are optional.