NEU Grand Library | TRNC Library


	NEU GRAND LIBRARY Opening Hours: Monday-Saturday, 08:00-20:00 \| E-mail: library@neu.edu.tr

You are not logged in

Show Basket

Home

Advanced Search

Back

Add to Basket

Brief display

MARC Display

Reserve

An emotional system with application to blood cell type identification. (Khashman, Adnan,)

Bibliographical information (record 268028)

An emotional system with application to blood cell type identification.
Author:	Khashman, Adnan, Search Author in Amazon Books
Publisher:	Sage Publications Ltd.,

Edition:	2012.
Classification:	QA76.87
URL:	http://library.neu.edu.tr:2048/login?url=http://dx.doi.org/10.1177/0142331210366640

Detailed notes

- Emotional processes and responses are biological and psychological processes in humans. The modelling of these biological processes in machines is considered a challenging and often controversial task in the field of bio-inspired computing. This is because of the complexity of emotional processes and rarity of available emotional computational models. However, research into possible mechanisms that allow emotional phenomena to influence cognitive and behavioural processes and how they can be implemented in robots has recently been gaining momentum with successful results in modelling emotions within systems. The motive for this trend is that future machines and robots will be capable of making better decisions if their cognition is enhanced by emotions. We present in this paper an emotional neural network based on an improved emotional back propagation (EmBP) learning algorithm. The modified algorithm, namely the iEmBP learning algorithm, models within its configuration three emotional responses: anxiety, confidence and impression. The proposed iEmBP-based emotional neural network will be implemented to a blood cell type identification problem, and its performance will be compared to two other neural networks. Experimental results show that the iEmBP-based emotional neural network has improved performance when compared to the EmBP-based network, and significantly outperforms the conventional back propagation (BP)-based neural networks in recognition results and time cost.