Personal achievement: Udacity Deep Learning certificate

I did it

Today I graduated from Udacity and their Deep Learning program. After a quite turbulent time the past months, it really feels as an achievement.
Spending a few hours every evening in learning and applying my prior knowledge I now consider myself having a better academic background to machine learning than before and am looking forward to be spending even more time in this area in future.


During the course I was able to reach milestone by milestone. Following lessons were notable:

  1. Proper introduction to error and loss functions, backward propagation for complex networks using partial derivatives and more basics
  2. Proper introduction to data science techniques, data exploration and various frameworks
  3. Project: Predicting Bike Sharing patterns using multilayer perceptrons written in plain pure python
  4. Proper introduction to PyTorch
  5. Introduction to CNNs and computer graphics
  6. Project: Predicting dog breeds by re-configuring imagenet and applying CNNs
  7. Natural language processing and analysis (Word2Vec etc.)
  8. Small project: Sentiment prediction of movie reviews
  9. Generation of natural language and text
  10. Project: Generating TV scripts (RNN)
  11. Generative adversarial networks and CycleGANs
  12. Project: Generating new realistic faces
  13. Deploying the models to AWS SageMaker (PyTorch framework)

Although I have had quite some experience in some of these areas like computer vision, it was nice to see some approaches in a more academic way than I am used to when doing such things during working hours.\

Heiko A. Weber
Heiko A. Weber

I do things.