Advanced Deep Learning - Week 2

Course starts soon..


We will start now with a quiz based on the first week material

You have 6 minutes to answer the quiz.

The quiz link:
Quiz Link
It will be copied in Mattermost and in the Zoom chat.


Coursera updated the course last week. It may be that you have a different version.

You may have to reset the deadlines or wait for the deadlines to end to get access to the new materials.

Quick check: who is seeing what?


I would like you to bring a project idea next week

Next week we will have a small talk (5/10 minutes) about your projects

What do I mean by project idea

1) A general theme (ex: chess engine, face classification)

2) A dataset (or a workflow on how to get the data)

3) A general guess about the architecture (CNN, RNN) and if possible a paper or some implementations

Where can you find all of this?


HuggingFace (NLP)

Tensorflow Model Garden - Community

Papers with Code

other resources are possible, too

Quiz (10 Minutes)

  1. What is the biggest advantage of CNN over traditional FFNN? And what it its biggest limitation?
  2. Pooling layers are often applied after convolution layers. What would you think it would happen if you switch their position, so that the image will pass through the pooling step first and then will the convolution be applied?
  3. Why is it often the case that the width and height of an image reduces and the third dimension (channel) increase when getting deeper into the network?


For the next week

  • Finish the second week of the course, check all the videos (including MobileNet, its architecture and EfficientNet)
  • Prepare your project idea
  • Try the assignment on Residual Networks
  • Work on the assignment on Transfer Learning with MobileNet

Homework Distribution

  • [' John', 'Kimani']
  • [' Atul Kumar', 'Yadav']
  • [' Bennet', 'Möller']
  • [' Foroogh', 'Gharibi Monfared']
  • [' Chetanya abhinav', 'Nagineni']
  • [' Modeus', 'Abdelnaby']
  • [' Prosper Kwabena ', 'Adjei']
  • [' Syed Usman', 'Farooq']
  • [' Suman', 'Singha']
  • [' Niko', 'Schmidt']