Breast histopathological image analysis for cancer diagnosis using computer tools have gained much attention in the past decade due to the development in computation power. In particular, deep learning-based algorithms which uses deep features are popularly explored for analysing breast histopathological images. However, there exists several challenges in developing computer tools such as heterogeneous characteristic of cancerous cells, illumination variation, color variation etc. Moreover, deep learning models are dependent on large annotated datasets. However, limited benchmark breast histopathological image datasets restricts the application of deep learning models. In this regard, the present paper aims at classification of breast histopathological images at 100x magnification into benign and malignant using deep learning models. Further, this paper demonstrates that data augmentation can improve the accuracy of deep learning models for classification of breast histopathological images. This paper also demonstrates that transferring the features of deep learning models learnt on general object class to and fine tuning it to classify breast histopathological images gives competitive results.