Practical: Fundamentals of deep learning

Administrative Information

Keywords

deep learning,model building,bias,

Learning Goals

• Tune hyper-parameters
• Investigate model bias

Expected Preparation

Lesson materials

• Practical (HTML)
• Practical (.ipynb)

Instructions for Teachers

This is a 2.5-hour practical, where students will work in teams of 3. The overall aim of this practical is to identify bias for target groups (as defined in the EC ALTAI [1]). This is applied to both the data and the algorithms. Students must identify and discuss any biases that might affect users of the model. Building trustworthy models involve deeper dives and discussion on not only how good your model is but where it's weaknesses lie, and being honest and upfront when presenting the model metrics.

To that end this practical will ask students to investigate data and model bias from a target group viewpoint, to discuss potential and metric-driven issues that may arise from this work.

The following section describes the overview of the tasks and the time allocation that students should aim to follow. The tasks below are linked to the sections that you will need to run in this notebook, some code is scaffolded, some is presented in it's entirety, and sometimes there is no code presented at all. For each task, there is a description listed. There are some sections that are not linked but need to be run, for example, the import section.

Data set:

Data sets for teaching ethical AI (Census data set)

Outline

Acknowledgements

The Human-Centered AI Masters programme was Co-Financed by the Connecting Europe Facility of the European Union Under Grant №CEF-TC-2020-1 Digital Skills 2020-EU-IA-0068.

Lesson plan on SURF

WikiWijs page