Basic Information

  • Course Code: EE 782
  • Course Name: Advanced Topics in Machine Learning
  • Course Offered In: 2023-‘24
  • Semester Season: Autumn
  • Instructors: Prof. Amit Sethi
  • Prerequisites: A basic ML course and familiarity with basic probability and python
  • Difficulty (1 being easy and 5 being tough): 2

Course Content

  1. Basic Neural Networks
  2. Convolutional Neural networks
  3. Introduction to RNNs and LSTMs
  4. Neural Networks for NLP
  5. Regularization
  6. Generative Modelling
  7. Loss Functions
  8. Few Shot Learning
  9. Semi Supervised Learning
  10. Graph Neural Networks
  11. Domain Adaptation
  12. Uncertainty Estimation

    Feedback on Lectures

Lectures were mostly easy to understand and easy to follow except some lectures where he derived the backpropagation for LSTM. Lectures are also recorded and some of the lectures are also available on NPTEL and YouTube so you can see the specific topic you were not able to understand from there. There were no tutorial sessions as such however some lectures were conducted by his PhD students which were also easy to follow.

Feedback on Evaluations

  1. Two Programming Assignments (28%) Initially each was going to be 10% however assignment 1 was 15% and assignment 2 was 13% Assignments were not that difficult as there were many resources on internet and sufficient time was given for each assignment
  2. Weekly Quiz on Safe (8%) There were around 10 quizzes and best 8 were taken each being 1%. These were just to give the incentive for attendance even if you scored 50% marks full marks were given, so yeah do attend classes
  3. Midsem & Endsem (20% and 28%) Exams were easy and most of the questions were directly from slides or a little bit thinking was required, also the exams were open notes so you could also print the slides
  4. Project (18%) : Group of 2 or single You have to select a research paper from the field of ML you like and implement that paper or try to make some changes and observe. You also have to prepare a paper in IEEE format of 2-3 pages to show the work you did. Deadline for this was after endsems

As you can see the total % for above is 102, there can be other bonus questions in midsem, endsem and assignments. Grading was absolute with AA above 90

Study Material and Resources

Slides were shared for every lecture and were enough for most of the topics however you may need to look some lectures again to have better understanding. Books were shared however they were not needed much.

“Deep Learning,” by Ian Goodfellow and Yoshua Bengio and Aaron Courville [www.deeplearningbook.org/]

“Dive into Deep Learning,” by Zhang, Lipton, Li, Smola [d2l.ai/d2l-en.pdf]

“Understanding Machine Learning: From Theory to Algorithms,” by Shai Shalev-Shwartz and Shai Ben-David [www.cs.huji.ac.il/w~shais/UnderstandingMachineLearning/]

Follow-up Courses

No courses as such but you can into ML research, build custom models and do some follow up projects

Final Takeaway

Going to class, doing assignments and project should be enough to get a good grade in this course and try to do the assignments on your own as it will be very useful for your own knowledge as well as your resume