Authors:
Petr Ježek
and
Lukáš Vařeka
Affiliation:
Department of Computer Science and Engineering, New Technologies for the Information Society, Faculty of Applied Sciences, University of West Bohemia, Plzeň and Czech Republic
Keyword(s):
EEG, ERP, Cloud, HDFS, Hadoop, Spark, Experiment, Infrastructure.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Software Development
;
Symbolic Systems
Abstract:
Current infrastructures for experimental data, results and computational tools make a shift from locally maintained solutions to remote cloud-based infrastructures. It brings a higher availability, sustainability and performance. However, specifics of different research areas require development of customized solutions for individual research domains. For example, electroencephalography and event-related potentials (EEG/ERP) use specific devices, data formats and machine learning workflows. As a solution, a cloud-based system for the EEG/ERP domain containing a distributed data storage, a signal processing method library and a client GUI is presented. The signal processing method library is used for training of classifiers and classifying the data in the cloud-based system controlled by the GUI. The presented system was tested using a machine learning workflow based on the data stored in the system. In the workflow, various classifiers were trained and their parameters stored into th
e system. Finally, testing data were classified using previously trained classifiers.
(More)