Tutorial on Machine Learning

Tutorial mainly prepared by: Ender Konukoglu, Christian F. Baumgartner, Krishna Chaitanya

In this practical session, participants will learn about basics of machine learning and deep learning, and gain hands on experience with applying available tools.

This tutorial consists of two parts: The first part covers basic Machine Learning concepts and the second part is devoted to introduction for Deep Learning.

Machine learning tutorial consists of two parts:

• Part I: is devoted to learning the basics that will include the following concepts:
  (a) Reading and visualizing data.
  (b) Supervised learning: toy classification and regression with linear models.
  (c) Supervised learning: toy classification and regression with non-linear models.
  

• Part II: participants will work on real-world data problems:
  (d) automatic diagnosis of Alzheimer’s disease from brain MRI and age using regression.
  

Deep learning tutorial contains four parts:

• Part I: Covers basics of the tensorflow library.
• Part II: Covers multilayer perceptrons for regression.
• Part III: In this part, you will learn about the convolution operation and how the weights of a convolution kernel can be learned to produce a desired output.
• Part IV: In the last part you will learn how convolution layers along with other types of layers can be combined to build and train a state-of-the-art medical segmentation network.

Running the material

• Download the material by running git clone https://github.com/eerdil/USZ_ML_workshop in command line.

• Download the data from the link https://www.dropbox.com/s/i9djv4si1uc7nff/acdc_preprocessed.hdf5?dl=0 and copy into deep_learning->data folder.

The instructions for setting up your computer to run the material are given in the "installation_instructions.txt"

Contact

For any queries regarding the tutorial please contact:

Ertunc Erdil (ertunc.erdil@vision.ee.ethz.ch)