Show/Hide Menu
Hide/Show Apps
Logout
Türkçe
Türkçe
Search
Search
Login
Login
OpenMETU
OpenMETU
About
About
Open Science Policy
Open Science Policy
Open Access Guideline
Open Access Guideline
Postgraduate Thesis Guideline
Postgraduate Thesis Guideline
Communities & Collections
Communities & Collections
Help
Help
Frequently Asked Questions
Frequently Asked Questions
Guides
Guides
Thesis submission
Thesis submission
MS without thesis term project submission
MS without thesis term project submission
Publication submission with DOI
Publication submission with DOI
Publication submission
Publication submission
Supporting Information
Supporting Information
General Information
General Information
Copyright, Embargo and License
Copyright, Embargo and License
Contact us
Contact us
Colour coding based novel data representation and lightweight convolutional neural network architecture for hierarchical anomaly detection on ehealth applications
Date
2020-10
Author
Yatbaz, Hakan Yekta
Metadata
Show full item record
This work is licensed under a
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License
.
Item Usage Stats
369
views
0
downloads
Cite This
eHealth is on its way to become an essential industry due to the advancements in information technology. Human activity recognition (HAR) is one of the most popular areas within the scope of eHealth, particularly with applications in anomaly detection. Although there are various studies on HAR, most of them propose complex models that are not compatible with portable devices and wearables due to their restricted computing capabilities. In this thesis, new data representation is presented along with a lightweight convolutional neural network (CNN) for this purpose. An anomaly detection framework is presented, which uses ECG data for heart effort prediction of daily life activities. The novel data representation approach and the proposed deep learning model are tested on the MHEALTH dataset with two different validation techniques for accuracy and three different complexity metrics. The results show that the proposed approaches can achieve up to 96.92% and 97.06% accuracy for HAR, and heart effort level with five fold cross validation. In addition, the models proposed for inertial data based and ECG based predictions have sizes of 0.89 MB and 1.97 MB and have a complexity of 0.06 and 1.04 Giga FLOPS, respectively.
Subject Keywords
eHealth
,
Human Activity Recognition
,
Lightweight
,
Deep Learning
URI
https://hdl.handle.net/11511/69236
Collections
Northern Cyprus Campus, Thesis
Suggestions
OpenMETU
Core
NETWORK INTRUSION DETECTION WITH A DEEP LEARNING APPROACH
Kültür, Ebru; Acar, Aybar Can; Department of Cybersecurity (2022-2-7)
With the rapid growth of the information technology in several areas, providing security of those systems has gained more importance. As a result of this development in information technology, the complexity of cyber-attacks has also significantly increased. Therefore, traditional security tools such as Signature-based Intrusion Detection Systems (SIDS) have become insufficient for detecting new attacks. Intrusion Detection Systems (IDS) are used to monitor network traffic and capture malicious traffic. Tra...
Multi-modal learning with generalizable nonlinear dimensionality reduction
Kaya, Semih; Vural, Elif; Department of Electrical and Electronics Engineering (2019)
Thanks to significant advancements in information technologies, people can acquire various types of data from the universe. This data may include multiple features in different domains. Widespread machine learning methods benefit from distinctive features of data to reach desired outputs. Numerous studies demonstrate that machine learning algorithms that make use of multi-modal representations of data have more potential than methods with single modal structure. This potential comes from the mutual agreemen...
Database security in private database clouds
Çınar, Onur; Yazıcı, Adnan; Baykal, Nazife; Department of Information Systems (2015)
Cloud computing and systems are very popular in today’s Information Technology and Systems and have grown extremely. Cloud computing helps organizations solve increasing IT costs such as their licenses,power consumption,physical protection,by providing better standardization,higher benefit,greater performance,and quick responses of information services. This thesis presents the study of ensuring database security in cloud computing, protecting data especially from internal attacks, and managing high volumes ...
Towards secure e-elections in Turkey: Requirements and principles
Cetinkaya, Orhan; Cetinkaya, Deniz (2007-04-13)
E-democracy is a necessity in this era of computers and information technology. E-voting is one of the most significant pails of e-democracy, which refers to the use of computers or computerized voting equipment to cast ballots in an election.
Evaluation Of A Hospital Information System In An International Context: Towards Implementing Pb-Ism In Turkey
Özkan Yıldırım, Sevgi; Baykal, Nazife (WILEY, 111 RIVER ST, HOBOKEN 07030-5774, NJ USA, 2006-12)
High quality information technology (IT) support is an integral factor in the success of today's health sector. However, the medical profession is considered to be low level investors in IT when compared to other sectors. Many international comparisons look at the expenditure spent on health IT as a basis for determining how effective the systems are. This prescriptive mainly cost/benefit approach is believed to be deficient as a sole factor in evaluation procedures. The research in this study supports the ...
Citation Formats
IEEE
ACM
APA
CHICAGO
MLA
BibTeX
H. Y. Yatbaz, “Colour coding based novel data representation and lightweight convolutional neural network architecture for hierarchical anomaly detection on ehealth applications,” M.S. - Master of Science, Middle East Technical University, 2020.