RT Book, Section
T1 A methodology for the development of soft sensors with Kafka-ML
A1 Chaves, Antonio
A1 Martín-Fernández, Cristian
A1 Llopis-Torres, Luis Manuel
A1 Soler-Castillo, Enrique
A1 Díaz-Rodríguez, Manuel
K1 Internet de los objetos
K1 Detectores
K1 Aprendizaje automático (Inteligencia artificial)
AB Advances in the Internet-of-Things (IoT) field have allowed a wide variety of devices to be connected and send information continuously to the Internet. Thanks to this increase in data communication, machine learning (ML) and data science have been able to be applied to analyze and extract valuable intelligence from the IoT. In this sense, the IoT has also contributed to improving the design and implementation of soft sensors. Soft sensors are used to predict features that are difficult to measure directly because the sensor to do so does not exist or is very expensive. IoT real-time monitoring can be used in conjunction with ML techniques to infer those parameters that are difficult to achieve with specific sensors. There exist methodologies for the development of soft sensors, but there is a lack of a common tool to support the design and implementation of them, covering the phases from model training to visualization of predictions. In this chapter, we present a methodology to support soft-sensor development based on Kafka-ML, an open-source framework to manage ML pipelines. Kafka-ML will allow researchers to develop, train, and validate ML models and visualize real-time predictions using streaming data. To demonstrate the viability of our proposal, we developed a soft sensor that predicts nitrate levels from river watersheds.
PB Springer
YR 2023
FD 2023
LK https://hdl.handle.net/10630/30035
UL https://hdl.handle.net/10630/30035
LA eng
NO Chaves, A.J., Martín, C., Torres, L.L., Soler, E., Díaz, M. (2023). A Methodology for the Development of Soft Sensors with Kafka-ML. In: Sharma, R., Jeon, G., Zhang, Y. (eds) Data Analytics for Internet of Things Infrastructure. Internet of Things. Springer, Cham. https://doi.org/10.1007/978-3-031-33808-3_17
NO Springer en las obras colectivas permite postprint con 24 meses de embargo
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 23 ene 2026