Abstract

Incoming TV is an Android App (http://bit.ly/1vt7avj) which learns about user’s video preferences and preloads the best videos in advance for a superior user experience – no spinning beach balls anymore!

Incoming’s prefetching approach is unique as most of the (machine) learning is performed on the phone leading to increased privacy, lower communication overhead and very low backend cost – and interesting research problems in ‘energy-constraint ML’.

This talk will briefly outline the overall architecture and objectives which are primarily driven by content provider which clash in interesting ways with network operators objectives (mainly believed, and less so in reality).

Incoming.TV has been downloaded approx. 200,000 times and used across the world leading to a trove of interesting measurements for which we are looking for collaborators to mine.

Bio: Max is currently a Sr. Principal Researcher at NICTA where he just returned from birthing Incoming Media, a provider of mobile video solutions for the content industry. He is actively involved in most of Future Internet research testbeds around the world. He was a Research Professor at WINLAB , Rutgers U. and the Founder of Semandex, an early content-networking company.