The Indian Ocean Dipole (IOD) is the strongest interannual variability in the tropical Indian Ocean in autumn. It will influence the climate in many parts of the world through atmospheric teleconnection. The current coupled climate model has very limited IOD forecasting skills, which are far lower than the forecasting skills of El Ni?o events in the tropical Pacific. Due to super capability of deep learning in processing data, we use the convolutional neural network (CNN) of the deep learning and the multi-layer perceptron (MLP) of the artificial neural network, respectively, to perform IOD prediction. In order to explore the forecasting capabilities of CNN, this article only uses three initial conditions in the boreal spring which has the low prediction skill, namely January- February-March (JFM) and February-March-April (FMA)), March-April-May (MAM), to forecast the Indian Ocean Dipole Index (DMI), East Pole Index (EIO), and West Pole Index (WIO) for the next seven months. The results show that the CNN model can make useful prediction for the DMI, EIO and WIO at least 6-month ahead. Compared with the current state-of-the-art general coupled model, the CNN model can significantly improve the prediction skills of DMI index and EIO index, but has little improvement for WIO prediction skill. The CNN model is able to predict the strong IOD events in 1994, 1997 and 2019 well for the lead time longer than 7 months. In general, because of the CNN is better than the traditional neural network MLP for the IOD prediction due its strong capability in capturing the spatial structure characteristics of the Indian Ocean SST.