Javid, Irfan (2023) Modified gated recurrent unit based on update-prime gate for heart disease prediction. Doctoral thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p IRFAN JAVID.pdf Download (1MB) | Preview
	Text (Copyright Declaration) IRFAN JAVID COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (547kB) | Request a copy
	Text (Full Text) IRFAN JAVID WATERMARK.pdf Restricted to Registered users only Download (3MB) | Request a copy

Abstract

There is a growing interest in accurate methods to detect heart disease since this is one of the leading causes of mortality worldwide. This interest has spurred advancements in various fields, including the application of deep learning algorithms for heart disease detection. One such algorithm is the Gated Recurrent Unit (GRU), which falls under the category of Recurrent Neural Network (RNN) that comprises an update gate and reset gate. GRU is considered one of the most efficient prediction approaches, particularly on time series datasets. However, when GRU was used to address heart disease prediction, it encountered three significant problems such as failure of data dimensionality reduction, slow convergence rate and high computational cost. Therefore, this research proposed a model named Chi-square Gated Recurrent Unit (χ2-GRU) to solve the dimensionality reduction issues. However, still, the χ2-GRU model has a slow convergence rate. This problem is tackled by enhancing the χ2-GRU model to Chi-square U-prime Gated Recurrent Unit (χ2-U/GRU) by dividing the update gate functionality into two parts. Finally, the

Actions (login required)

View Item