Sensor Sequential Data-Stream Classification Using Deep Gated Hybrid Architecture

Sensors are the main components to supply information for resource management in a smart city. This paper studies the sensor data-stream classification problem using different time series state-of-the-art classification models. In this study, we found that the hybrid architecture of gated recurrent units and temporal fully convolutional neural network (GRU-FCN) model outperforms the existing state-of-the-art classification techniques in most of the benchmark sensor-obtained datasets. Moreover, the GRU-FCN model is simpler than the other existing gate-based recurrent classification architectures. Thus, it is an appropriate model to be implemented on small or portable hardware devices.

[1]  Mark Kretschmar,et al.  CHAPTER 8 – Capacitive and Inductive Displacement Sensors , 2005 .

[2]  Yixin Chen,et al.  Multi-Scale Convolutional Neural Networks for Time Series Classification , 2016, ArXiv.

[3]  Jason Lines,et al.  Time series classification with ensembles of elastic distance measures , 2015, Data Mining and Knowledge Discovery.

[4]  Janez Demsar,et al.  Statistical Comparisons of Classifiers over Multiple Data Sets , 2006, J. Mach. Learn. Res..

[5]  Olufemi A. Omitaomu,et al.  Weighted dynamic time warping for time series classification , 2011, Pattern Recognit..

[6]  Jason Lines,et al.  Time-Series Classification with COTE: The Collective of Transformation-Based Ensembles , 2015, IEEE Transactions on Knowledge and Data Engineering.

[7]  U. Berardi,et al.  Smart Cities: Definitions, Dimensions, Performance, and Initiatives , 2015 .

[8]  T. Pohlert.  The Pairwise Multiple Comparison of Mean Ranks Package (PMCMR) , 2016 .

[9]  Annalisa Cocchia.  Smart and Digital City: A Systematic Literature Review , 2014 .

[10]  Tim Oates,et al.  Time series classification from scratch with deep neural networks: A strong baseline , 2016, 2017 International Joint Conference on Neural Networks (IJCNN).

[11]  A. Vanolo.  Smartmentality: The Smart City as Disciplinary Strategy , 2014 .

[12]  Houshang Darabi,et al.  LSTM Fully Convolutional Networks for Time Series Classification , 2017, IEEE Access.

[13]  Yoshua Bengio,et al.  Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling , 2014, ArXiv.

[14]  Margarita Angelidou,et al.  Smart city policies: A spatial approach , 2014 .

[15]  Magdy A. Bayoumi,et al.  Deep Gated Recurrent and Convolutional Network Hybrid Model for Univariate Time Series Classification , 2018, International Journal of Advanced Computer Science and Applications.

[16]  Patrick Schäfer.  The BOSS is concerned with time series classification in the presence of noise , 2014, Data Mining and Knowledge Discovery.