Deep Learning-based VGGNet, GoogleNet, and DenseNet121 Models for Cervical Cancer Prediction

Cervical cancer is still a serious health issue in the world and underscores the importance of developing accurate, effective diagnostic tools. This research investigates cervical cancer detection by using the three latest deep learning models, DenseNet121, VGGNet and Google Net. The evaluation refers to the analysis of accuracy of models under various learning rates as 0.01 and 0.001, VGGNet model was found to be superior with an accuracy of 98.93%. Compared to the above figures, DenseNet121 and GoogleNet achieved accuracies of 93.21% and 93.67%, respectively. This result highlights VGGNet's better ability to provide accurate cervical cancer detection under this learning rate. When an adjustment was made as regards to the value of learning rate by setting it at 0.001, on the other hand, DenseNet121 and VGGNet exhibited accuracies of 93.27% and 89.99% respectively, whereas GoogleNet resulted it as 97.33%. These results imply that the choice of learning rate greatly impacts deep learning models' performance in cervical cancer detection. The VGGNet model outperformed with a higher learning rate while GoogleNet demonstrated its strength compared to lower rates. Study shows that it is possible to optimize deep learning models for cervical cancer detection thus improving diagnostic accuracy and ultimately, patients' outcomes.