Condition monitoring of rolling element bearings in rotating machines is important for preventing breakdowns of industrial machinery. This presentation will propose a new two-stage method. This will introduce compressive sampling to reduce significantly the amount of data required for this. Furthermore, it will introduce deep neural networks for feature selection (using unsupervised learning based on sparse autoencoder) and classification of bearing conditions. This is a two stage method. The experimental results show that the proposed method is able to achieve higher levels of accuracy even with extremely compressed measurements compared with existing techniques. This method can be found in a recently published article in Mechanical Systems and Signal Processing (Ahmed, Wong, Nandi, 2018, 99:459-477).