Control and Optimization
Soghra Mikaeyl Nejad
Abstract
Gene expression signatures reflect the response of cell tissues to diseases, genetic disorders, and drug treatments, containing hidden patterns that can provide valuable insights for biological research and cancer diagnostics. This studyproposes a hybrid ...
Read More
Gene expression signatures reflect the response of cell tissues to diseases, genetic disorders, and drug treatments, containing hidden patterns that can provide valuable insights for biological research and cancer diagnostics. This studyproposes a hybrid deep learning approach combining convolutional neural networks (CNNs) and support vector machines (SVMs) to classify cancer types using unstructured gene expression data. We applied three hybrid CNN-SVM models to a dataset of 10,340 samples spanning 33 cancer types from the Cancer Genome Atlas. The CNN component extracted latent features from the gene expression data, while the SVM replaced the softmax layer to enhance classification robustness. Among the proposed models, the Hybrid-CNN-SVM model achieved superior performance, demonstrating excellent prediction accuracy and outperforming other models. This study highlights the potential of hybrid deep learning frameworks for cancer type prediction and underscores their applicability to high-dimensional genomic datasets.
