Porting Rulex Software to the Raspberry Pi for Machine Learning Applications on the Edge

Abstract

Edge Computing enables to perform measurement and cognitive decisions outside a central server by performing data storage, manipulation, and processing on the Internet of Things (IoT) node. Also, Artificial Intelligence (AI) and Machine Learning applications have become a rudimentary procedure in virtually every industrial or preliminary system. Consequently, the Raspberry Pi is adopted, which is a low-cost computing platform that is profitably applied in the field of IoT. As for the software part, among the plethora of Machine Learning (ML) paradigms reported in the literature, we identified Rulex, as a good ML platform, suitable to be implemented on the Raspberry Pi. In this paper, we present the porting of the Rulex ML platform on the board to perform ML forecasts in an IoT setup. Specifically, we explain the porting Rulex’s libraries on Windows 32 Bits, Ubuntu 64 Bits, and Raspbian 32 Bits. Therefore, with the aim of carrying out an in-depth verification of the application possibilities, we propose to perform forecasts on five unrelated datasets from five different applications, having varying sizes in terms of the number of records, skewness, and dimensionality. These include a small Urban Classification dataset, three larger datasets concerning Human Activity detection, a Biomedical dataset related to mental state, and a Vehicle Activity Recognition dataset. The overall accuracies for the forecasts performed are: 84.13%, 99.29% (for SVM), 95.47% (for SVM), and 95.27% (For KNN) respectively. Finally, an image-based gender classification dataset is employed to perform image classification on the Edge. Moreover, a novel image pre-processing Algorithm was developed that converts images into Time-series by relying on statistical contour-based detection techniques. Even though the dataset contains inconsistent and random images, in terms of subjects and settings, Rulex achieves an overall accuracy of 96.47% while competing with the literature which is dominated by forward-facing and mugshot images. Additionally, power consumption for the Raspberry Pi in a Client/Server setup was compared with an HP laptop, where the board takes more time, but consumes less energy for the same ML task.

Incremental and parallel proximal SVM algorithm tailored on the Jetson Nano for the ImageNet challenge

“…Because of tremendous growth in manufacturing powerful, low-cost embedded devices, the edge computing becomes a popular choice for machine learning and IoT projects (Ajani et al , 2021; Koul et al , 2019; Kulkarni et al , 2020; Kurniawan, 2021; Latif et al , 2021; Mazzia et al , 2020; Norris, 2020; Pooyandeh and Sohn, 2021; Rahmaniar and Hernawan, 2021; Taylor et al , 2018). The health applications, computer vision and deep learning are tailored on the Jetson Nano (Black, 2022; Budek, 2021; Franklin, 2019; Mishra and Devleker, 2021; Mittal, 2019; Mohan et al , 2021; Rehman et al , 2021; Zualkernan et al , 2022) and the Raspberry Pi (Daher et al , 2021; Glegola et al , 2021; Iodice, 2018).…”

A multithreaded CUDA and OpenMP based power‐aware programming framework for multi‐node GPU systems

“…In the future, benchmarking and extending the solution towards other types of GPUs will be of interest, specifically mobile and embedded type of compute devices in multi-node systems. It has been demonstrated that such devices offer less power demanding computing 40,41 and exploration of performance-power trade-offs using power capping might result in non-trivial configurations under more strict power limitations as compared to powerful GPUs in the traditional servers and HPC systems. Additionally, extending the implementation to use available CPU cores for computations will be performed, according to the concept presented in Appendix E.…”

ImageNet classification with Raspberry Pis: federated learning algorithms of local classifiers

“…Due to the tremendous growth in the manufacturing of powerful and low-cost embedded devices, edge computing has become a popular choice for machine learning and IoT projects (Koul et al , 2019; Kulkarni et al , 2020; Norris, 2020). Health applications, computer vision and deep learning are being tailored to run on the Jetson Nano (Rehman et al , 2021; Zualkernan et al , 2022), and the multiclass proximal SVM algorithm has been optimized for the Jetson Nano (Do, 2022) as well as the Raspberry Pi (Daher et al , 2021; Glegola et al , 2021).…”