The prostate cancer is the second leading cancer among men. The incident rates of prostate cancer are increasing in, all over the world. The histopathology based grading system, is mainly used in prostate cancer diagnosis, in which the pathologist assigns a Gleason grade, based on the architecture of the tissue. It is important in risk assessment and treatment planning for the patients. Cancer affects the epithelial cell of prostate tissue. Several computer aided methods are proposed in the literature, using various handcrafted image features and machine learning algorithms, for the classification of grades. In this paper, we review various automated methods which are used in the prostate cancer grading. In the Deep Learning models the image features where automatically extracted from the images but in conventional machine learning approaches, the features are manually selected. The deep learning papers in the histopathology of prostate cancer are limited since the development of deep learning models have started in recent past. The conventional machine learning models perform with high accuracy in small datasets while the deep learning models perform with high accuracy in adequate datasets. For automating the cancer grading deep learning models using Convolutional Neural Networks (CNN) are performing well while compared with the conventional machine learning models.