Teaching
Course materials and resources for classes taught. This page is based on the structure of the "Intro to ML" course and contains a draft syllabus, schedule, and placeholders for slides, assignments and code.
This page collects course materials, lecture notes, assignments, and links for courses I’ve taught. The content below is a draft based on the “Intro to ML” course structure; some materials (slides, datasets, GitHub) are left as placeholders where the original site requires access.
Course: Intro to Machine Learning (drafted from IITGN “Intro to ML” site) Course: Intro to Machine Learning (IITGN short course — exact details below)
Course summary: An introductory short course that builds concepts from fundamentals and equips students with intuition and mathematics behind ML algorithms. Students will implement classical ML techniques and gain hands-on experience.
Pre-requisites:
- Basic Linear Algebra (MA103 level is sufficient)
- Basic Probability
- Familiarity with Python, NumPy, and pandas. Knowledge of PyTorch is preferable but not required.
Learning objectives:
- Learn mathematical foundations (SVD, matrix calculus, probability distributions) used in ML
- Implement and understand unsupervised methods (clustering, dimensionality reduction)
- Implement and understand supervised methods (linear/logistic regression, decision trees, ensembles)
- Understand neural network basics including MLPs and convolutional neural networks
- Apply regularization, optimization, and model-selection techniques (cross-validation, hyperparameter tuning)
Day-wise lecture schedule (as in the course PDF):
- 31/10/2025 — Mathematical Foundations for ML
- SVD
- Matrix calculus
- Probability distributions
- Time: 21:30–22:50 — Instructor: Ridham Patel
- 04/11/2025 — Unsupervised Learning (Clustering)
- K-means, hierarchical clustering
- Dimensionality reduction; curse of dimensionality
- Parametric vs non-parametric methods
- Matrix factorisation (if time permits)
- Time: 21:30–22:50 — Instructor: Ridham Patel
- 06/11/2025 — Decision Trees and Ensembles
- Tree construction: splitting, entropy, information gain
- Overfitting/underfitting in trees
- Random forests intuition; feature importance; hands-on tree algorithms
- Time: 21:30–22:50 — Instructor: Aryan Solanki
- 10/11/2025 — Linear Regression & Mathematical Intuition
- Simple linear regression: fitting and prediction
- Geometric perspective, cost functions
- Variants (e.g., polynomial regression)
- Logistic regression and logits; bias–variance tradeoff
- Time: 21:30–22:50 — Instructor: Ridham Patel
- 11/11/2025 — MLPs & Gradient Descent
- Perceptron, Multi-layer Perceptron (MLP)
- Activation functions: ReLU, Sigmoid, Tanh
- Probability interpretation; gradient descent and backpropagation
- Time: 21:30–22:50 — Instructor: Aryan Solanki
- 13/11/2025 — Autograd and Regularisation
- Autograd systems: computational graphs and automated chain rule
- Intuition for regularization; Ridge and LASSO
- Hyperparameter tuning and cross-validation (including k-fold)
- Time: 21:30–22:50 — Instructor: Romit Mohane
- 14/11/2025 — Convolutional Neural Networks
- Convolution operation: intuition and mathematics
- Convolutional layers, pooling, strides, padding
- Modern architectures overview (LeNet, ResNet concepts)
- Time: 21:30–22:50 — Instructor: Romit Mohane
General instructions (from PDF):
- Additional reading material will be provided for interested readers with every lecture.
- Code will be provided with every lecture/topic where applicable; students are encouraged to implement the code themselves to gain hands-on experience.
Registration: Closing date/time in the PDF — 30th October 2025, 17:00.
Course materials:
- Lecture slides, code, and additional readings will be linked here when available. The original short-course PDF is available as an attachment.
Attendance: This short course was attended by 200+ students — the largest short course ever held at IITGN.
If you want, I can:
- Add a downloadable copy of the PDF into the site’s
assetsand link it here. - Convert this single page into a
_teaching/collection with one markdown per lecture (useful for per-lecture notes, slides, and code). - Add direct links to slides/code if you provide the URLs or files.