PV115 Knowledge discovery lab
Themata
- Analysis of experiments done in spring (Framework = Outllier filtering + feature
selection + Classifier, for about 60 datasets)
- Extension of the framework and new experiments
neural networks MG, BŠ
gradient boosting MB
- OpenML KŠ
- Medical data JK
Schedule of seminar presentations
In addition to seminars, consultations of the projects, take place
once per two weeks
Oct 20 Luboš Popelínský, Outlier filtering to increase performance. Part 1
Oct 27 Consultations (LP, Outlier filtering to increase performance. Part 2 moved)
Nov 3 Katka Švecová
Nov 10 Introduction of the projects (everybody but KŠ)
Nov 24, Dec 1 Consultations
Dec 8 Dušan Hetlerovič
Dec 15 Katka Švecová, Tool for processing OD+FS+CLF results
Jan 5 J. Kloss, M. Bankovič
Jan 12 B. Ševc, M. Geletka
Evaluation
To succeed, it is necessary to finish a project, including
- a presentation = slides+video (typically 2 per term)
- a short report (explaining details, if needed, of the project)