Pregled bibliografske jedinice broj: 222699
Inducing Decision Trees from Reduced Datasets
Inducing Decision Trees from Reduced Datasets // Annals of DAAAM for 2005 & Proceedings of the 16th International DAAAM Symposium "Intelligent Manufacturing & Automation: Focus on Young Researches and Scientists" / Katalinić, Branko (ur.).
Beč: DAAAM International Vienna, 2005. str. 357-358 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 222699 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Inducing Decision Trees from Reduced Datasets
Autori
Štajduhar, Ivan ; Stojković, Nino
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
Annals of DAAAM for 2005 & Proceedings of the 16th International DAAAM Symposium "Intelligent Manufacturing & Automation: Focus on Young Researches and Scientists"
/ Katalinić, Branko - Beč : DAAAM International Vienna, 2005, 357-358
Skup
The 16th International DAAAM Symposium "Intelligent Manufacturing & Automation: Focus on Young Researches and Scientists"
Mjesto i datum
Opatija, Hrvatska, 19.10.2005. - 22.10.2005
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
Dataset reduction; Decision tree; Sample exclusion
Sažetak
Datasets often contain large quantities of redundant and unnecessary samples, that only harden the problem of learning classification models that describe them. What if one could learn a classification model from a reduced dataset, that would be similar or almost equal to the one gained from the complete dataset? Problem of such exclusion of samples from a dataset for building equivalent decision trees is presented in this paper. Selecting the adequate samples from a dataset is very time consuming. Several experiments have shown that only heuristic methods are capable of completing that task. An efficient heuristic method of sample exclusion is proposed.
Izvorni jezik
Engleski
Znanstvena područja
Elektrotehnika, Računarstvo
