Welcome to the Introduction to Deep Learning course offered in WS18.
Lecture
Thursdays (18:00-20:00) - HOERSAAL MI HS 1 (00.02.001)
Lecturers: Prof. Dr. Laura Leal-Taixé and Prof. Dr. Matthias Niessner.
ECTS: 6
SWS: 4
Tutorial
Tuesdays (18:00-20:00) - Interims Hörsaal 1 (5620.01.101)
Tutors: Patrick Dendorfer and Andreas Rössler
For details about the exercises please read the README.md, which is distributed with the exercise. In particular, you will submit your solutions via our submission website.
Content
- 18.10 - Lecture 1: Introduction to the lecture, Deep Learning, Machine Learning.
- 25.10 - Lecture 2: Machine Learning Basics: Linear regression, Classification and Loss Functions.
- 08.11 - Lecture 3: Introduction to neural networks
- 15.11 - Lecture 4: Backpropagation
- 22.11 - Lecture 5: Optimization I
- 29.11 - Lecture 6: Optimization II
- 06.12 - Lecture 7: Training Neural Networks Part I: Regularization, Activation functions, Weight initialization, Gradient flow, Batch normalization and Hyperparameter optimization.
- 13.12 - Lecture 8: Training Neural Networks Part II: Parameter updates, Ensembles and Dropout.
- 10.01 - Lecture 9: Convolutional Neural Networks (CNN) I
- 17.01 - Lecture 10: CNN II: common architectures, VGG, ResNet, Inception
- 24.01 - Lecture 11: Recurrent networks (RNN), LSTM
- 31.01 - Guest lecture: Oriol Vinyals form Google deepmind
Slides
You can now download the slides in PDF format:
- Lecture 1, Lecture 2 , Lecture 3, Lecture 4, Lecture 5, Lecture 6, Lecture 7, Lecture 8 , Lecture 9, Lecture 10, Lecture 11
Prerequisites
- Strong mathematical background: linear algebra, calculus.
- Previous knowledge of Python is needed.
- Machine Learning background is recommended.
- Previous knowledge of PyTorch is recommended.
Moodle
We use Moodle for discussions and to distribute important information. Please check the News and Discussion boards regularly or subscribe to them.
The slides and all material will also be posted on Moodle.
Exercise and tutorial schedule
- 23.10 - Organisation Exercises (Bonus) and Math Background I (attendency recommended)
- 30.10 - Math Background II
- 06.11 - Introduction to Exercise system and Exercise 0
- 20.11 - Introduction to Exercise 1
- 04.12 - Solution to Exercise 1 and Introduction to Exercise 2
- 18.12 - Solution to Exercise 2 and Introduction to Pytorch and Exercise 3
- 22.01 - Solution to Exercise 3 and Introduction to Exercise 4
- 05.02 - Solution to Exercise 4
- TBA - Additional pratical session
EXERCISE 0:
- Topics: Setup and test the submission system
EXERCISE 1:
- Topics: Softmax classifier & Two-layer NN
- Starting date: 20.11.2018
- Due date: 3.12.2018, 23:59
EXERCISE 2:
- Topics: Fully Connected Neural Network
- Starting date: 4.12.2018
- Due date: 17.12.2018, 23:59
EXERCISE 3:
- Topics: Classification with PyTorch and Image Segmentation
- Starting date: 8.1.2019
- Due date: 22.1.2019, 17:59
EXERCISE 4:
- Topics: Facial Keypoint detection and RNN MNIST classification
- Starting date: 22.1.2019
- Due date: 4.2.2019, 23:59
FINAL EXAM:
- Exam date: 28.02.2019, 16h-17:30h (the date is fixed and cannot be changed)
If you have any questions regarding the organization of the course, do not hesitate to contact us at: i2dl@dvl.in.tum.de
For questions on the syllabus, exercises or any other questions on the content of the lecture, we will use the Moodle discussion board.
Office hours
In addition to the tutorial we also offer to discuss questions in person in our office hours at room 02.09.051:
Office hours start in the second week of lectures
- Monday: 12:15 - 13:45 & 15:00 - 16:30
- Tuesday: 10:00 - 11:30 & 14:00 - 15:30
- Wednesday: 16:00 - 17:30
- Thursday: 10:30 - 12:00 & 14:00 - 15:30
- Friday: 11:00 - 12:30 & 13:00 - 14:30
External (non-TUM) Students
We want to provide access to our lecture for as many students as possible. Therefore, we ask external students that are not TUM students and do not have access to TUMonline to register to Moodle and send us their student information via email. We will then add you to our Moodle course where you will find addtional information and all the course material.
