Breast cancer remains one of the top causes of death for women and many efforts have been made in the form of screening programs for prevention. Computer-assisted detection has become essential due to the programs’ continuous increase in the quantity of mammograms that collect. However, the design of computer-aided detection methods to enhance diagnosis without requiring many systematic measurements has not yielded appreciable gains in performance metrics. In this case, the application of deep learning-derived automatic image processing methods offers a potentially useful approach to assist in breast cancer diagnosis. In this work, we describe a technique using deep learning for multi-class breast cancer classification based on a Convolutional Neural Network (CNN) model. The proposed technique is to classify cancers of the breast as more than just benign or malignant; instead, we forecast the subtype of the tumors, such as lobular fibroadenoma, carcinoma, etc. The effectiveness of test results using a MobileNet CNN mode with breast cancer CSV dataset is used for evaluation. In the multi-class breast tumor classification task, the proposed technique outperformed state-of-the-art algorithms.