Following the COVID-related shelter-at-home orders, Comcast’s network experienced a significant increase in demand for bandwidth. Though an unpreceded situation, Comcast was well prepared to meet this demand due to a multitude of efforts relating to how we manage our network capacity. One factor in our increased downstream and upstream capacity is Octave, a machine learning platform that makes daily configurations to the network to optimize performance through management of DOCSIS profiles. In this talk, you’ll learn about the main concepts behind Octave and how the system came together in the context of the pandemic.
Back to All Events
Earlier Event: October 28
"Real world reinforcement learning" by John Langford of Microsoft
Later Event: October 28
"ForestFlow.ai: An LFAI incubating ML server" by Ahmad Alkilani of DreamWorks