Data Stream Mining with Swarm Decision Table in Fog Computing Environment
Jiaxue Li, Simon James Fong, Tengyue Li, Wei Song · 2018
Fog computing, as an expansion of Cloud computing, provides edge intelligence where data mining will be implemented. Compared with big data computation at the Cloud platform, distributed Fog nodes process data generated by Internet of Things (IoT) sensors directly at the edge of network. Fog computing can not only relieve heavy workload at the Cloud server, but also increase the speed of data analytics locally. However, faced with continuous data stream, the Fog node should be capable of real-time data mining with high accuracy and lightweight as well. In this paper, a combination of feature selection methods coupled with swarm intelligence and decision Table classifier called Swarm Decision Table (SDT) are proposed. SDT is designed to find appropriate data mining model in the Fog computing environment. Based on a scenario of chemical gas sensors, a simulation experiment will be carried out to evaluate the performance of different swarm feature selection algorithms with decision Table model. The results revealed that the SDT model with the right feature selection method is suitable for Fog computing node, in terms of speed and accuracy.