Redesign and Evaluation of a Teaching Tool for Autoencoders

Theoretical (Analytical):

Practical (Implementation):

Literature Work:

Overview

Algorithms get more and more complex. Computers can visualize their execution "live." Thus, we can teach algorithms "live" with computers. Task: Redesign and evaluate a teaching tool for Autoencoders.

Problem Statement

Autoencoders are a popular type of neural network. However, teaching them to students can be difficult. There is a need for a well-designed and evaluated teaching tool for Autoencoders.

Tasks

  • Analyze the viability of an existing teaching tool for autoencoders.
  • Redesign and improve the existing teaching tool.
  • Plan an evaluation of the developed framework.

Requirements

Good programming skills in Java/Python and TypeScript/JavaScript. Good familiarity with the D3 library.

Scope/Duration/Start

  • Scope: Bachelor or Master
  • 3 Month Project, 6 Month Thesis
  • Start: immediately

Contact

References

Johannes Fuchs, Petra Isenberg, Anastasia Bezerianos, Matthias Miller, Daniel A. Keim. EduClust - A Visualization Application for Teaching Clustering Algorithms. Eurographics (Education Papers) 2019: 9-16.