LOW COMPLEXITY RECURRENT NEURAL NETWORKS FOR EDGE COMPUTING

Oleh Sinkevych

Abstract


This paper is dedicated to the development of recurrent neural networks in order to supplement the AI-based devices like microcontrollers and other mist computing systems. Due to the insignificant computational power of the edge deviсes the aim of the study is to design and analyze low complexity sequence models for a basic sensory time series forecasting on an example of univariate indoor temperature data. The description of data preparation and transformation followed by the models configuration via different architectures like simple LSTM and GRU is provided. To calculate an optimal set of hyper-parameters for the multiple neural network architectures a genetic algorithm has been implemented. The results of numerical experiments conducted for each model configuration consisting of both unidirectional and bidirectional cell connections are discussed. In addition to these studies the scheme of deploying the developed low-complexity models on STM32 microcontroller joined with the high-performance hub is proposed.

Key words: edge computing, recurrent neural networks, time series, genetic algorithm.


Full Text:

PDF

References


[1] Deep Learning for Edge Computing Applications: A State-of-the-Art Survey / [F. Wang, M. Zhang, X. Wang та ін.]. // IEEE Access. – 2020. – №8. – P. 58322 – 58336. DOI: 10.1109/ACCESS.2020.2982411.

[2] Edge computing use case examples [Електронний ресурс]. – 2020. – Режим доступу до ресурсу: https://stlpartners.com/edge_computing/10-edge-computing-use-case-examples/.

[3] Mao J. Application of learning algorithms in smart home IoT system security / J. Mao, Q. Lin, J. Bian. // Mathematical Foundations of Computing. – 2018. – №1. – P. 63–76. DOI: 10.3934/mfc.2018004.

[4] Smart Home System Based on Deep Learning Algorithm / Y. Peng, J. Peng, J. Li, L. Yu. // Journal of Physics: Conference Series. – 2019. – №1187. – P. 032086. DOI: 10.1088/1742-6596/1187/3/032086.

[5] Design of a Prototype Neural Network for Smart Homes and Energy Efficiency / T.Teich, F. Roessler, D. Kretz, S. Franke. // Procedia Engineering. – 2014. – №69. – P. 603–608. DOI: 10.1016/j.proeng.2014.03.032.

[6] Chakraborty T. Home automation using edge computing and Internet of Things / T. Chakraborty, S. Datta. // IEEE International Symposium on Consumer Electronics (ISCE). – 2017. – P. 47–49. DOI: 10.1109/ISCE.2017.8355544.

[7] Design of Smart Home System Based on Collaborative Edge Computing and Cloud Computing / Q.Ma, H. Huang, W. Zhang, M. Hua. // Algorithms and Architectures for Parallel Processing. – 2020. – P. 355–366. DOI: 10.1007/978-3-030-60248-2_24.

[8] Self-Learning Algorithm to Predict Indoor Temperature and Cooling Demand from Smart WiFi Thermostat in a Residential Building / [K. Huang, K. Hallinan, R. Lou та ін.]. // Sustainability. – 2020. – №12. – P. 7110–7124. DOI: https://doi.org/10.3390/su12177110.

[9] Deep Learning architecture for temperature forecasting in an IoT LoRa based system / [I. Ouahab, B. Abdelhakim, A. Astito та ін.]. // Conference: the 2nd International Conference on Networking, Information Systems & Security (NISS19). – 2019. – P. 1–6. DOI: 10.1145/3320326.3320375.

[10] AI expansion pack for STM32CubeMX [Електронний ресурс] – Режим доступу до ресурсу: https://www.st.com/en/embedded-software/x-cube-ai.html.

[11] Jin H. Auto-Keras: An Efficient Neural Architecture Search System / H. Jin, Q. Song, X. Hu. // ACM SIGKDD International Conference. – 2019. – P. 1946–1956. DOI: 10.1145/3292500.3330648.

[12] Embedding Sequence Model in STM32 Based Neuro-Controller / [O. Sinkevych, Y. Boyko, O. Rechynskyi та ін.]. // 2021 IEEE 12th International Conference on Electronics and Information Technologies (ELIT). – 2021. – P. 113–118. DOI: 10.1109/ELIT53502.2021.9501132.

[13] Siami-Namini S. The Performance of LSTM and BiLSTM in Forecasting Time Series / S. Siami-Namini, N. Tavakoli, A. Siami Namin. // 2019 IEEE International Conference on Big Data (Big Data). – 2019. – P. 3285–3292. DOI: https://doi.org/10.1109/BigData47090.2019.9005997.




DOI: http://dx.doi.org/10.30970/eli.16.2

Refbacks

  • There are currently no refbacks.