Towards a global and abstract end-to-end architecture for data analysis and transformation with ML/DL Application cases: Medical IoT and IoHT
Hayat Zaydi, Zohra Bakkoury · 2021 International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME) · 2021
Nowadays, data is a valuable resource for businesses in all industries. However, getting value out of it requires a sophisticated process resulting in life cycle of data analysis, covering data generation and creation, capture, storage, preprocessing, and the choice of algorithms and techniques to be applied according to many criteria, right up to the cost-effectiveness of results through deployment with end-user. The application of this abstract concept concretely takes the form of a range of architectures and infrastructures for processing, storing, and analysing data from connected objects or data lakes. In this work, we propose a general and abstract model of an end-to-end architecture for data processing and transformation from connected objects and data assets of concerned organizations, This data-oriented workflow allows exploiting on-premises infrastructures and services offered by cloud world with load balancing and bypassing shortcomings and limitations of Cloud. Then we instantiate this model for a healthcare data use case with ML/DL.