Tuesday, January 12th 2016
Prof.dr. C.N. Schizas, University of Cyprus
Variable target values Neural Network for dealing with imbalanced datasets
A classification algorithm is proposed for dealing with highly imbalanced datasets that often appear in biomedical problems. Its idea comes from the way a neural network is trained in order to get a decent hypothesis out of a dataset that comprises a huge sized majority class and a tiny size minority class. This situation is especially probable when forming machine learning databases describing rare medical conditions. The algorithm is tested on a large dataset for predicting the risk of preeclampsia in pregnant women and preliminary results are encouraging. Conventional machine learning algorithms tend to provide poor hypothesis for extremely imbalanced datasets by favoring the majority class. The proposed algorithm is not trained on the basis of the mean squared error objective function and thus avoids the overwhelming effect of the highly asymmetric class sizes.
Colloquium coordinators are Prof.dr. M. Aiello (e-mail :
Prof.dr. M. Biehl (e-mail:
University of Groningen and University of Oldenburg (UOL) celebrate 40 years of collaboration in 2020. On 29 October 2020 (Festakt-date) the UG and the UOL will officially renew the partnership that they started 40 years ago. The anniversary will...
The COVID-19 restrictions have been tightened once again. To many staff members and students, this feels like a step backwards. We chatted with members of the Board of the University Jouke de Vries, Cisca Wijmenga and Hans Biemans about managerial...
Biological clocks have sizeable effects on the performance of elite athletes. This conclusion was drawn by chronobiologists from the University of Groningen after studying the times achieved by swimmers in four different Olympic Games. Shifting the...