Hybrid arithmetic optimization algorithm with deep transfer learning based microarray gene expression classification model

• Published in: International journal of information technology (Singapore. Online) Vol. 16; no. 6; pp. 3923 - 3928
• Main Authors: Gowri, B. Shyamala, Nair, S. Anu H., Kumar, K. P. Sanal
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.08.2024; Springer Nature B.V
• Subjects: Algorithms; Arithmetic; Artificial Intelligence; Cancer; Classification; Computer Imaging; Computer Science; Disease; Exploitation; Extraction procedures; Feature extraction; Gene expression; Image Processing and Computer Vision; Machine Learning; Optimization; Original Research; Pattern Recognition and Graphics; Software Engineering; Vision; Median Filtering; Gene Expression; Hyperparameter Tuning; Microarray; Deep Transfer Learning
• ISSN: 2511-2104; 2511-2112
• DOI: 10.1007/s41870-024-01901-2

Microarray gene expression (MGE) data classification is a vital challenge in biomedical and genomics research, designed to understand the difficult relations among genes and numerous biological states. In this procedure, high-dimensional gene expression data is primarily pre-processed and converted to create it responsive to machine learning (ML) techniques. Feature selection and dimensionality reduction models are frequently used to remove the most appropriate gene signatures. medicine, and drug improvement, making it a vital element of recent genomics studies. Therefore, this manuscript presents a hybrid arithmetic optimization algorithm with a deep transfer learning-based MGE data classification (HAOADTL-MGEDC) model. The HAOADTL-MGEDC technique initially undergoes the conversion of gene expression data into image format. Then, the preprocessing of the images takes place using the median filtering-based noise removal and contrast enhancement approach. Besides, the feature extraction procedure is implemented by the usage of the capsule network (CapsNet) model. Meanwhile, the CapsNet hyperparameters can be optimally selected by the use of HAOA. Finally, the parallel bidirectional gated recurrent unit (BiGRU) methodology has been used for the classification of GE data. The experimental values of the HAOADTL-MGEDC technique have been tested on a benchmark dataset. Extensive simulation analysis inferred the higher of the HAOADTL-MGEDC technique compared to other DL approaches.