Two distributed semi-supervised algorithms for mining of complex networks using Giraph

Heterogeneous networks are large graphs consisting of different types of nodes and edges. They are an important category of complex networks, but the process of knowledge extraction and relations discovery from these networks are so complicated and time-consuming. Moreover, the scale of these networ...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:bioRxiv
Hauptverfasser: Erfan Farhangi Maleki, Ghadiri, Nasser, Maryam Lotfi Shahreza, Maleki, Zeinab
Format: Paper
Sprache:Englisch
Veröffentlicht: Cold Spring Harbor Cold Spring Harbor Laboratory Press 15.11.2019
Cold Spring Harbor Laboratory
Ausgabe:1.2
Schlagworte:
Algorithms
Bioinformatics
Statistical analysis
Apache Giraph
Complex Networks
Heterogeneous Label Propagation
Drug Repositioning
Distributed Graph Processing
Semi-Supervised Learning
ISSN:2692-8205, 2692-8205
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Heterogeneous networks are large graphs consisting of different types of nodes and edges. They are an important category of complex networks, but the process of knowledge extraction and relations discovery from these networks are so complicated and time-consuming. Moreover, the scale of these networks is steadily increasing. Thus, scalable and accurate methods are required for efficient knowledge extraction. In this paper, two distributed label propagation algorithms, namely DHLP-1 and DHLP-2, in the heterogeneous networks have been introduced. The Apache Giraph platform is employed which provides a vertex-centric programming model for designing and running distributed graph algorithms. Complex heterogeneous networks have many examples in the real world and are widely used today for modeling complicated processes. Biological networks are one of such networks. As a case study, we have measured the efficiency of our proposed DHLP-1 and DHLP-2 algorithms on a biological network consisting of drugs, diseases, and targets. The subject we have studied in this network is drug repositioning, aimed at saving both time and cost by suggesting new indications for the current drugs. We compared the proposed algorithms with similar non-distributed versions of them namely MINProp and Heter-LP. The experiments revealed that the runtime of the algorithms has decreased in the distributed versions rather than non-distributed ones dramatically. The effectiveness of our proposed algorithms against other algorithms is supported through statistical analysis of 10-fold cross-validation as well as experimental analysis. Footnotes * The title is modified to reflect the paper content more properly.
Bibliographie:SourceType-Working Papers-1
ObjectType-Working Paper/Pre-Print-1
content type line 50
ISSN:2692-8205
2692-8205
DOI:10.1101/477463