| Category | Difficulty |
|---|---|
| HW | 5 |
| Quiz | 5 |
Do not let the name fool you, it is a very demanding course. Introduction to Deep Learning is one of the most well run class in CMU. Prof. Bhiksha has fine-tuned the course over multiple iterations. Roughly about 200 students take the course every semester. Piazza is well handled with the average response time being under 5 mins. TAs are also very helpful.
The course structure is very dynamic and keeps evolving with every offering. Some of the main topics covered are
- Multi-layer Perceptron
- Back propagation and training the Neural Network
- Regularization
- Convolutional Neural Networks
- Recurrent Neural Networks
- GRU and LSTM
- Sequence-to-Sequence models
- Hopfield Networks
- Boltzmann Machines
- Variational Autoencoders
- Generative Adversarial Networks (GANs)
- Reinforcement Learning
There are 4 assignments. Each assignment will have 2 parts. Part 1 is typically something that needs to be submitted on autolab. Part 2 is in-class Kaggle Competition.
There will be one quiz per week - 14 quizzes throughout the semester. The 12 best performances are used for calculating the grade.
The graduate level course has a project component as well which counts towards 25% of the grade.
Attend classes regularly and start assignments early! Assignments take a lot of time to complete. Many potential areas to get stuck in assignments. So start early and attend TA office hours. Not backlogging it to the final few weeks is the key. There will be regular TA office hours. So allocate a lot of time to finishing assignments.
The quiz will be released on Friday night and it is due on Sunday night. Quizzes are very interesting but they take considerable amount of time.
Regarding the project, there will be an initial project proposal which is due in the first month. TA mentors will be allocated who will guide the teams on projects. There is a midterm review and a final poster presentation. The project encourages students to explore some topics in detail and go beyond what is covered in class.
It is easy to fall behind after the first month. So attend classes regularly start assignments early. Allocate enough time for extra readings required for completing the quiz and assignments. Recitations are very useful to get started with assignments.