Digitalni repozitorij raziskovalnih organizacij Slovenije

Iskanje po repozitoriju
A+ | A- | Pomoč | SLO | ENG

Iskalni niz: išči po
išči po
išči po
išči po

Možnosti:
  Ponastavi


Iskalni niz: "avtor" (Sašo Džeroski) .

1 - 6 / 6
Na začetekNa prejšnjo stran1Na naslednjo stranNa konec
1.
Comparing algorithm selection approaches on black-box optimization problems
Ana Kostovska, Anja Janković, Diederick Vermetten, Sašo Džeroski, Tome Eftimov, 2023, objavljeni znanstveni prispevek na konferenci

Objavljeno v DiRROS: 25.03.2024; Ogledov: 29; Prenosov: 12
.pdf Celotno besedilo (582,18 KB)

2.
MsGEN : measuring generalization of nutrient value prediction across different recipe datasets
Gordana Ispirova, Tome Eftimov, Sašo Džeroski, Barbara Koroušić-Seljak, 2023, izvirni znanstveni članek

Povzetek: In this study, we estimate the generalization of the performance of previously proposed predictive models for nutrient value prediction across different recipe datasets. For this purpose, we introduce a quantitative indicator that determines the level of generalization of using the developed predictive model for new unseen data not presented in the training process. On a predefined corpus of recipe embeddings from six publicly available recipe datasets (i.e., projecting them in the same meta-feature vector space), we train predictive models on one of the six recipe datasets and test the models on the rest of the datasets. In parallel, we define and calculate generalizability indexes which are numbers that indicate how generalizable a predictive model is i.e., how well will a predictive model learned on one dataset perform on another one not involved in the training. The evaluation results prove the validity of these indexes – their relation with the accuracy of the predictions. Further, we define three sampling techniques for selecting representative data instances that will cover all parts from the feature space uniformly (involving data from all datasets) and further will improve the generalization of a predictive model. We train predictive models with these generalized datasets and test them on instances from the six recipe datasets that are not selected and included in the generalized datasets. The results from the evaluation of these predictive models show improvement compared to the results from the predictive models trained on one recipe dataset and tested on the others separately.
Ključne besede: ML pipeline, predictive modeling, nutrient prediction, recipe datasets
Objavljeno v DiRROS: 25.09.2023; Ogledov: 358; Prenosov: 167
.pdf Celotno besedilo (3,27 MB)
Gradivo ima več datotek! Več...

3.
Algorithm instance footprint : separating easily solvable and challenging problem instances
Ana Nikolikj, Sašo Džeroski, Mario Andrés Muñoz, Carola Doerr, Peter Korošec, Tome Eftimov, 2023, objavljeni znanstveni prispevek na konferenci

Ključne besede: black-box optimization, algorithms, problem instances, machine learning
Objavljeno v DiRROS: 15.09.2023; Ogledov: 255; Prenosov: 169
.pdf Celotno besedilo (2,03 MB)
Gradivo ima več datotek! Več...

4.
On the influence of aging on classification performance in the visual EEG oddball paradigm using statistical and temporal features
Nina Omejc, Manca Peskar, Aleksandar Miladinović, Voyko Kavcic, Sašo Džeroski, Uroš Marušič, 2023, izvirni znanstveni članek

Povzetek: The utilization of a non-invasive electroencephalogram (EEG) as an input sensor is a common approach in the field of the brain–computer interfaces (BCI). However, the collected EEG data pose many challenges, one of which may be the age-related variability of event-related potentials (ERPs), which are often used as primary EEG BCI signal features. To assess the potential effects of aging, a sample of 27 young and 43 older healthy individuals participated in a visual oddball study, in which they passively viewed frequent stimuli among randomly occurring rare stimuli while being recorded with a 32-channel EEG set. Two types of EEG datasets were created to train the classifiers, one consisting of amplitude and spectral features in time and another with extracted time-independent statistical ERP features. Among the nine classifiers tested, linear classifiers performed best. Furthermore, we show that classification performance differs between dataset types. When temporal features were used, maximum individuals’ performance scores were higher, had lower variance, and were less affected overall by within-class differences such as age. Finally, we found that the effect of aging on classification performance depends on the classifier and its internal feature ranking. Accordingly, performance will differ if the model favors features with large within-class differences. With this in mind, care must be taken in feature extraction and selection to find the correct features and consequently avoid potential age-related performance degradation in practice.
Ključne besede: aging, elderly, machine learning, visual oddball study, brain-computer interface
Objavljeno v DiRROS: 01.02.2023; Ogledov: 378; Prenosov: 202
.pdf Celotno besedilo (3,50 MB)
Gradivo ima več datotek! Več...

5.
Uporaba metod strojnega učenja za preučevanje odnosov med značilnostmi branik in okoljem
Jernej Jevšenak, Sašo Džeroski, Tom Levanič, 2017, izvirni znanstveni članek

Povzetek: Različne študije so pokazale, da lahko z nelinearnimi metodami bolje opišemo (modeliramo) odnos med branikami in okoljem. V naši študiji smo primerjali (multiplo) linearno regresijo (MLR) in štiri nelinearne metode strojnega učenja: modelna drevesa (MT), ansambel bagging modelnih dreves (BMT), umetne nevronske mreže (ANN) in metodo naključnih gozdov (RF). Za primerjavo teh metod modeliranja smo uporabili štiri množice podatkov. Natančnost naučenih modelov smo ocenili z metodo 10-kratnega prečnega preverjanja (ang. 10-fold cross-validation) na naši množici in preverjanjem na dodatni testni množici. Na vseh množicah smo dobili boljše statistične kazalce za nelinearne metode s področja strojnega učenja, s katerimi lahko pojasnimo večji delež variance oz. dobimo manjšo napako. Nobena metoda se ni pokazala kot najboljša v vseh primerih, zato je smiselno predhodno primerjati več različnih metod in nato uporabiti najprimernejšo, npr. za rekonstrukcijo klime.
Ključne besede: strojno učenje, primerjava metod, dendroklimatologija, umetne nevronske mreže, modelna drevesa, ansambel modelnih dreves, naključni gozdovi, linearna regresija
Objavljeno v DiRROS: 21.02.2018; Ogledov: 5234; Prenosov: 3270
.pdf Celotno besedilo (1,18 MB)
Gradivo ima več datotek! Več...

6.
Windthrow factors - a case study on Pokljuka
Nikica Ogris, Sašo Džeroski, Maja Jurc, 2004, izvirni znanstveni članek

Povzetek: This paper presents a case study in windthrow. The case study area was 1.7 ha of two forest gaps on the Pokljuka plateau, Slovenia, where strong wind had blown down 44 trees. An additional 44 standing trees closest to the fallen trees were used as a control group for comparative purposes. The following variables were measured for fallen trees: breast diameter, height, crown diameter and height as well, the number and diameter of roots, the volume of the root system, and root rot. Standing trees were measured for breast diameter, height, crown diameter and height, and the number and diameter of roots. The data were analysed using the machine learning methods in the Weka computer program. The most important factors of windthrow in the case study area were: storm wind (speed above 17 m/s), wet shallow soil, and the edges ofthe forest gaps. The results of the case study show that breast diameter, tree height and the presence of root rot can be classified as windthrow factors.
Ključne besede: wind, windthrow, root rot, factors of windthrow
Objavljeno v DiRROS: 12.07.2017; Ogledov: 4111; Prenosov: 1854
.pdf Celotno besedilo (1,44 MB)

Iskanje izvedeno v 0.17 sek.
Na vrh