Worldwide breast cancer is an important health concern and predominant disease found among women, affecting millions of lives every year. This disease can occur when uncontrolled cells are formed in the breast and lead to breast cancer. Fortunately, Breast cancer is neither an infection nor transmissible disease. With an early detection of this disease can ensure better and advanced treatment to reduce the mortality rate. Early detection using breast self-exams regularly, mammogram screening can improve treatment outcomes to reduce the number of cases. Radiologists recommend mammograms for diagnosing and screening the breast conditions, aiming to detect benign and malignant abnormalities. Deep Learning techniques have shown high accuracy in classification of breast cancer. Now-a-days, Deep Learning approaches have produced out performing accuracies in the medical image analysis field. By leveraging large datasets of mammograms, deep learning algorithms can automatically extract complex features and patterns from images like benign and malignant lesions with accuracy of high and also using deep learning models can be continuously trained and improved with new data, enhancing their performance. In this research work we considered the two datasets CBIS-DDSM (Curated Breast Imaging Subset of Digital Database for Screening Mammography) dataset and MIAS (Mammographic Image Analysis Society) dataset with CLAHE used for improving the contrast of the image, AlexNet CNN (Convolution Neural Network) model used to find the mammographic input images into benign and malignant and achieved the accuracy of 96% and 96.2% respectively.