Akademska digitalna zbirka SLovenije - logo
VSE knjižnice (vzajemna bibliografsko-kataložna baza podatkov COBIB.SI)
  • Prediction sample size in data mining tasks using additive incremental approach
    Brumen, Boštjan ; Jaakkola, Hannu ; Welzer-Družovec, Tatjana
    Data mining has received an increasing amount of attention lately. The fact that data hide important knowledge is a primary motivating factor. Most of the research is focusing on data mining in large ... databases. The identified problem is how to predict the behavior of a learning algorithm based on the measurements obtained early in the data mining phase. We present the idea of an additive incremental approach. The approach will be used to predict the accuracy of a learning algorithm, in our case a classifier, for a large number of samples based on a small one, without actually running the algorithm. Our contribution is a presentation of preliminary ideas of a doctoral study conducted so far. The results presented in this paper confirm our approach, but the formal verification and the full explanation will be a part of our future work.
    Vir: Information modelling and knowledge bases XII (Str. 264-270)
    Vrsta gradiva - članek, sestavni del
    Leto - 2001
    Jezik - angleški
    COBISS.SI-ID - 7285270

Vnos na polico