Analysis of Hypertuning Based CNN Breast Cancer Predictor for Different Random States

Breast cancer is one type of cancer that specifically targets the cells in the breast tissue. Millions of women worldwide are killed by this type of cancer, which is among the most common. A effective treatment strategy and improved patient outcomes depend on early identification of breast cancer. In order to predict and detect breast cancer before it spreads, this study employs CNN. To train and assess the proposed model, 569 breast cancer patient records from the Wisconsin Diagnostic dataset are employed. The dataset is separated between training and testing sets using 80:20 ratios with 0, 1, and 2 random states. Using random state 1, benign and malignant cases are split in the same proportion as in the original dataset. The binary cross-entropy function is used to calculate the discrepancy between the actual class label and the predicted class probability. In this study, the loss function is lowered using an Adam optimizer. Our test results demonstrate that the proposed method may accurately diagnose and forecast breast cancer.