15:30 - 16:00
Patterns that matter
Prof. dr. Arno Siebes (UU)

Patterns, or local models, a far more detailled insight in the data than the better known global models. Unfortunately, the number of patterns tends to explode: more patterns than data is not uncommon. In recent years we have developed the Krimp algorithm that, based on a MDL criterion, selects a relatively small set of patterns that together describe the data very well. Over the years we have used Krimp for a wide variety of data mining problems. In this talk I will focus on two, viz., data generation and the imputation of missing values.



Back to symposium program