Online Data Stream Analytics for Dynamic Environments Using Self-Regularized Learning Framework

For the upcoming Internet of things era, a plethora of data will need to be processed on a real-time basis. However, most prior data streaming-related studies ignored the semantic meaning of the input data, leading to unexpected or even unacceptable contextual models. This issue may worsen given the concept-drift problem in a dynamic environment. Since streaming data do not have ground-truth labels, it is challenging to reflect upon what has been learned for continuous improvement in learning performance. To address the aforementioned issues, this study incorporates self-regulated learning from pedagogy. There are three elements in self-regulated learning: Forethought, Performance, and Reflection. Each element's corresponding module is designed and implemented in this study. The goal is to reduce the need for human intervention in online data streaming analytics, while enhancing the abilities of semantic understanding and metacognitive self-monitoring. With the aforementioned enhancements, the experimental results of our enhanced model reached a level comparable to those of prior studies that required human interventions. Furthermore, the proposed approach helps guarantee that the model does not go astray from a preset learning goal.