The temporal distance between events conveys information essential for many time series tasks such as speech recognition and rhythm detection. While traditional mod-els such as hidden Markov models (HMMs) and discrete symbolic grammars tend to dis-card such information, recurrent neural networks (RNNs) can in principle le to make use of it. As an advanced variant of RNNs, long short-term memory (LSTM) has an alta-tive (arguably better) mechanism for bridging long time lags. We propose a couple of deep neural network-based models to detect abnormal start-ups, unusual CPU and memory con-sumptions of the application processes running on smart phones. Experiment results showed that the proposed neural networks achieve remarkable performance at some reason-able computational cost. The speed advantage of neural networks makes them even more competitive for the applications requiring real-time response, offering the proposed models the potential for practical systems.