Optimizing Convolutional Neural Networks for Cancer Biomarker Identification in Genomic Data: Challenges and Future Directions

Abstrak

Genomic analysis has become a major focus in cancer research to identify biomarkers that are important for more precise diagnosis and therapy. However, a major challenge in genomic analysis is the complexity and high dimensionality of genomic data, which requires sophisticated analysis approaches. This study aims to develop a deep learning model based on Convolutional Neural Networks (CNNs) that can recognize cancer biomarker patterns from genomic data with high accuracy. Relevant genomic data were collected and processed, then used to train CNNs models using optimization and regularization techniques. The CNNs model was then evaluated using validation data to measure its performance. The evaluation results show that although the model has improved in reducing the loss value, the accuracy obtained is still not optimal. The model is not fully able to identify cancer biomarker patterns accurately from the available genomic data. This research provides an important foundation for further development in genomic data analysis using deep learning. Suggestions for further research include the use of more representative data, optimization of model architecture, data augmentation, regularization, and external validation to improve model performance in cancer biomarker identification.