Mathematical Foundations of ML Course | IISc Bangalore | Prof. Prathosh A P
Course Details
| Exam Registration | 1965 |
|---|---|
| Course Status | Ongoing |
| Course Type | Elective |
| Language | English |
| Duration | 12 weeks |
| Categories | Computer Science and Engineering, Artificial Intelligence, Data Science |
| Credit Points | 3 |
| Level | Undergraduate/Postgraduate |
| Start Date | 19 Jan 2026 |
| End Date | 10 Apr 2026 |
| Enrollment Ends | 02 Feb 2026 |
| Exam Registration Ends | 20 Feb 2026 |
| Exam Date | 25 Apr 2026 IST |
| NCrF Level | 4.5 — 8.0 |
Master the Math Behind the Magic: A Deep Dive into the Mathematical Foundations of Machine Learning
Machine Learning (ML) has transformed from an academic curiosity to the driving force behind modern technology. Yet, for many aspiring practitioners, the field can feel like a black box—powerful but inscrutable. The key to unlocking true mastery lies not in blindly using libraries, but in understanding the rigorous mathematical principles that make ML work. A new, comprehensive course offered by the Indian Institute of Science (IISc) Bangalore, taught by the distinguished Prof. Prathosh A P, is designed to provide exactly that foundational understanding.
About the Course Instructor: Prof. Prathosh A P
The course is led by an exceptional academic and industry expert. Prof. Prathosh A P earned his Ph.D. from IISc Bangalore in 2015 in temporal data analysis, remarkably just three years after completing his B.Tech in 2011, and has authored several top-tier publications. His career includes impactful tenures at corporate research labs like Xerox Research India and Philips Research, where he focused on healthcare analytics and generated 15 commercialized U.S. patents.
In 2017, he joined IIT Delhi as an Assistant Professor before returning to IISc Bangalore as a faculty member in the Department of Electrical Communication Engineering. His research delves into deep representational learning, cross-domain generalization, and signal processing for vision and speech. Beyond academia, he is the co-founder of Cogniable.Tech, a healthcare AI start-up that won the Government of India AI Start-up Challenge, and actively collaborates with institutions like AIIMS. Prof. Prathosh uniquely bridges cutting-edge AI with deep explorations in Sanskrit and Indian philosophical sciences, offering a rare perspective on the field.
Course Overview: Building from the Ground Up
This 12-week course is meticulously structured for senior undergraduates and postgraduate students in EECS disciplines. It assumes a basic background in probability theory, linear algebra, and Python programming.
The journey begins with core concepts like Empirical Risk Minimization and Bayes Optimality, establishing the fundamental "why" behind learning algorithms. It then systematically builds up to both classical and modern paradigms:
- Classical ML: Linear models, kernel machines (SVMs), decision trees, and ensemble methods.
- Deep Learning: A thorough exploration of MLPs, CNNs, RNNs, and the revolutionary Transformer architecture.
- Unsupervised & Probabilistic Learning: Clustering, PCA, the EM algorithm, and an introduction to generative models like GANs and VAEs.
The course emphasizes the crucial link between theory and practice, with coding assignments designed to translate mathematical concepts into functional ML applications.
Detailed 12-Week Course Layout
| Week | Topics Covered |
|---|---|
| Week 1 | Introduction to Learning Paradigms; Empirical Risk Minimization |
| Week 2 | Bayes Optimality; Density Estimation via Divergence Minimization |
| Week 3 | Maximum Likelihood & MAP Estimates; Non-Parametric Methods (k-NN, Parzen Window) |
| Week 4 | Linear Models: Regression, Least Squares, Logistic Regression |
| Week 5 | Regularization & Generalization: Bias-Variance, Ridge, Lasso |
| Week 6 | Kernel Machines & SVMs: Maximum Margin, Duality, Kernel Trick |
| Week 7 | Neural Networks: Perceptron, Gradient-Based Optimization, Backpropagation |
| Week 8 | Convolutional Neural Networks (CNNs): Architectures, Transfer Learning |
| Week 9 | Sequence Models: RNNs, GRUs, LSTMs, Backpropagation Through Time |
| Week 10 | Attention & Transformers: Self-Attention, Encoder-Decoder Architecture |
| Week 11 | Ensembles & Decision Trees: Random Forests, Boosting (AdaBoost, XGBoost) |
| Week 12 | Unsupervised Learning: Clustering, EM Algorithm, PCA; Preview of GANs & VAEs |
Essential Reading & Industry Relevance
The course draws from authoritative texts in the field, ensuring a deep and rigorous learning experience. Key references include:
- Pattern Recognition and Machine Learning by Christopher Bishop
- Probabilistic Machine Learning series by Kevin P. Murphy
- Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville
- Understanding Machine Learning: From Theory to Algorithms by Shalev-Shwartz and Ben-David
This course is highly relevant for industry support, with direct applications sought by major IT and tech companies globally, including Google, Microsoft, Amazon, IBM, Oracle, Infosys, Accenture, and GE. A strong grasp of these mathematical foundations is invaluable for roles in research, development, and applied AI/ML.
Who Should Take This Course?
This course is ideal for students and professionals who want to move beyond being users of ML toolkits to becoming innovators and problem-solvers. If you aim to:
- Understand the "why" behind algorithm choices and hyperparameters,
- Read and comprehend advanced ML research papers,
- Build robust, efficient, and novel ML models from first principles, or
- Pursue a career in AI research or advanced development,
then mastering these mathematical foundations is your critical next step. Under the guidance of Prof. Prathosh A P—a scholar who embodies the synergy of deep technical expertise, successful innovation, and philosophical inquiry—this course offers a unique and powerful pathway to true proficiency in the science of machine learning.
Enroll Now →