Overview
The course provides the basics required to understand the architecture and function of Convolutional Neural Networks (CNN). Based on example data, participants study concrete classification problems, present CNN as a solution method and analyze its functionality as applied to the problems in question. What is more, participants will have the opportunity to work on a set problem or a problem of their own choosing - in groups an supervised by one of the lecturers.
Learning targets
The participants
- know the basic terms and concepts of image-based object classification and their validation methods;
- know typical examples suitable for the application of a deep learning approach;
- understand the structure of multi-layered neural networks and their functionality and understand their implementation;
- know the essential components of a neural network and their representation and parameterization in a Deep Learning framework (TensorFlow, Keras);
- can appropriately configure and parameterize a model to address existing and new problems in the field of image data classification.
In order to obtain two ETCSs, participants are required to write and submit a report in which their set/chosen problem is analyzed and discussed.