With the huge ascent of the populace, breast cancer gives a high mortality rate. Breast cancer is identified as the most harmful and risky malignancy that has recently been recognized. An automated cancer detection framework assists clinical practitioners to diagnose cancer risk early and provides dependable, effective, and speedy intervention so that the risk of mortality is greatly decreased. This paper presents breast cancer detection and classification based on deep convolutional neural network (DCNN) feature extraction, and further segmentation using SqueezeNet (SqN) is used to find the severity feature of cancer mass. Initially, the SqN is utilized to classify the cancer mass into either normal or abnormal class of mammogram. Here, the SqN is tuned by the Fractional War Strategy Optimization (FrWSO), which is the grouping of Fractional Calculus (FC). The exactness, responsiveness, accuracy, negative prescient worth, fake negative rate, false positive rate, F1 score, and correlation coefficient are helpful to classify the cancer mass. The implementation shows that proposed DCNNs have the best precision, accuracy, and F1 score of 95.57%, 96.82%, and 96.9%, individually.