Multi Layered Unsupervised Machine Learning for Detection of Real Time Network Service Impairment (2023)
By Eric Frishman, Comcast; Devangna Kaushish, Comcast; Russell Harlin, Comcast; Aditya Vallabhajosyula, Comcast; Eswaramoorthy Subramaniam, Comcast
Finding and categorizing network impairments is a significant challenge. Using Machine Learning (ML) to study network traffic patterns, we can quickly discover places where behaviors depart from statistical norms and highlight them. As the divergences may stem from common underlying characteristics themselves, ML modeling can further separate and classify situations to speed in diagnoses and resolution, leading to improved customer experience and satisfaction as well as more efficient use of staff resources. In this paper and presentation, we will walk through project and process flow, and ML considerations.
By clicking the "Download Paper" button, you are agreeing to our terms and conditions.