Training process visualisation

From Master Projects
Jump to: navigation, search


About Training process visualisation

  • This project has not yet been fulfilled.
  • This project fits in the following Bachelor programs: {{#arraymap:|, |xXx|bachelorproject within::xXx|,}}
  • This project fits in the following masterareas: {{#arraymap:Knowledge Technology and Intelligent Internet Applications, Computational Intelligence and Selforganisation, Technical Artificial Intelligence, AI and Communication, Multimedia, Software Engineering, Information and Communication Technology, Computer Science and Communication, Information Sciences|, |xXx|project within::xXx|,}}


Description

In machine learning, training a model is a process that has several steps. In each step, the fit of your model on the data at hand is supposed to improve, so gradually you'll approach an optimal minimum value for the error function (ideally). One can perhaps use a simple progress bar to show progress, but it's far from ideal. This project offers quite some freedom (see below), which is also a challenge, which should be kept in mind. The sub-tasks (roughly) are the following: research literature for visualisation and other methods that might be relevant; choose a direction (picture? video? audio?); select algorithms for which the visualisation is best suited; come up with your own way of process visualisation, given your previous choices; implement it; find a way to evaluate it; evaluate it; write it up in a thesis. Ideally, students interested should have at least some knowledge of machine learning or data mining (perhaps evolutionary computation), visualisation or related subjects. If you are interested, please let me (Zoli) know.