Abstract Deep learning has been one of the most prominent machine learning techniques nowadays, being the state-of-the-art on a broad range of applications where automatic feature extraction is needed. Many such applications also demand varying costs for different types of mis-classification errors, but it is not clear whether or how such cost information can be incorporated into deep learning to improve performance. In this work, we first design a novel loss function that embeds the cost information for the training stage of cost-sensitive deep learning. We then show that the loss function can also be integrated into the pre-training stage to conduct costaware feature extraction more effectively. Extensive experimental results justify the validity of the novel loss function for making existing deep learning models cost-sensitive, and demonstrate that our proposed model with cost-aware pre-training and training outperforms non-deep models and other deep models that digest the cost information in other stages.