Xiaofeng Xie, Di Wu, Siping Liu and Renfa Li are with the Key Laboratory for Embedded and Networking Computing of Hunan Province, Hunan University. Di Wu is the corresponding author (Email: dwu@hnu.edu.cn). IoT Data Analytics Using Deep Learning Xiaofeng Xie, Di Wu, Siping Liu, Renfa Li Abstract: Deep learning is a popular machine learning approach which has achieved a lot of progress in all traditional machine learning areas. Internet of thing (IoT) and Smart City deployments are generating large amounts of time-series sensor data in need of analysis. Applying deep learning to these domains has been an important topic of research. The Long-Short Term Memory (LSTM) network has been proven to be well suited for dealing with and predicting important events with long intervals and delays in the time series. LTSM networks have the ability to maintain long-term memory. In an LTSM network, a stacked LSTM hidden layer also makes it possible to learn a high level temporal feature without the need of any fine tuning and preprocessing which would be required by other techniques. In this paper, we construct a long-short term memory (LSTM) recurrent neural network structure, use the normal time series training set to build the prediction model. And then we use the predicted error from the prediction model to construct a Gaussian naive Bayes model to detect whether the original sample is abnormal. This method is called LSTM-Gauss-NBayes for short. We use three real-world data sets, each of which involve long-term time-dependence or short-term time-dependence, even very weak time dependence. The experimental results show that LSTM-Gauss-NBayes is an effective and robust model. 1. Introduction In many areas, such as the natural sciences, social sciences, and engineering, phenomena are best analyzed with time-series data. In the field of Internet of things, time series data can be generated by weather stations, RFID tags, IT infrastructure components, and some other sensors, as shown in Fig. 1(a). In each business process and every application of the Internet of things sensor, the time series data can be used for process optimization or knowledge discovery. When carefully analyzed such data can reveal operational trends, patterns, variability, changes, covariation, cycle abnormalities, anomaly and abnormal value rate. Traditional time series processing technique uses a statistical indicator such as cumulative sum (CUSUM) and exponentially weighted moving average (EWMA) in a time window [1] to detect potential changes in the distribution. The length of the time window usually needs to be predetermined and the result is highly dependent on the parameter. In addition, there are some sequence models, such as conditional random field model, d Kalman filter, Markov model, dealing with sequential data but are ill-equipped to learn long-range dependencies. What’s more, other models require domain knowledge or feature engineering, thus they provide fewer opportunities for accidental discovery. In contrast, neural network learning techniques allow unforeseen structures to be found. Recurrent neural network (RNN) [2] can theoretically solve long delayed tasks without requiring predefined time steps. However, because of the simplicity of the hidden layer units’ structure, gradient explosion or vanishing gradient [3] is easy to occur over longer time series tasks. The long short-term memory (LSTM) neural network [4] is a variant of the
