izpis_h1_title_alt

Konstruktivna indukcija s samokodirniki in gručenjem
ID Kuhar, Yannick (Author), ID Robnik Šikonja, Marko (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (2,63 MB)
MD5: A4CD30D2E70787344D69F0CFCDAF93DD

Abstract
Časovna kompleksnost algoritmov za gručenje je odvisna od dimenzionalnosti vhodnih podatkov, zato so počasni na visokodimenzionalnih podatkih. Problem bomo rešili z globokim samokodirnikom. Uporabili smo ga za kompresijo podatkov v manj dimenzij s čimmanjšo izgubo informacije. Reimplementirali in razširili smo postopek DeepCluster, ki so ga predlagali Tian et al [26]. Izvorno ogrodje podpira le algoritma za gručenje K-voditeljev in GMM. Razširili smo ga s hierarhičnim gručenjem, algoritmom DBSCAN in ansambelskim gručenjem. Ocenili smo kvaliteto gruč in samokodirnik interpretirali s konstruktivno indukcijo. Originalni in razširjeni postopek se v naših poskusih nista izkazala za uspešna, smo pa s konstruktivno indukcijo vizualizirali znanje modela in ga predstavili na razumljivejši način.

Language:Slovenian
Keywords:gručenje, konstruktivna indukcija, interpretabilnost modelov, samokodirniki
Work type:Bachelor thesis/paper
Typology:2.11 - Undergraduate Thesis
Organization:FRI - Faculty of Computer and Information Science
Year:2020
PID:20.500.12556/RUL-116753 This link opens in a new window
COBISS.SI-ID:32331523 This link opens in a new window
Publication date in RUL:08.06.2020
Views:978
Downloads:211
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Constructive induction using autoencoders and clustering
Abstract:
The time complexity of most clustering algorithms depends on the dimensionality of the input data and thus most clustering algorithms are slow on highdimensional data. To solve this problem, we trained a deep autoencoder and used it to compress the input data into a lower dimensional space with information loss. We reimplemented and extended the DeepCluster framework proposed by Tian et al [26]. The original framework supports only K-means and GMM clusterings. We extended it with hierarchical clustering, DBSCAN, and ensemble clustering. We evaluated the clusters and interpreted the autoencoder with constructive induction. Both frameworks proved to be unsuccessful in our experiments. However, we were able to interpret the model and visualize its knowledge

Keywords:clustering, constructive induction, model interpretability, autoencoders

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back