Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Fraternali concertation june25bruxelles


Published on

Published in: Technology, Travel
  • Be the first to comment

Fraternali concertation june25bruxelles

  1. 1. Content exploitation: men and machines DG Connect -Unit G1 – Converging Media and Content Bruxelles, June 16 2014 Piero Fraternali, Politecnico di Milano
  2. 2. What content? Corporate content • Google is indexing the earth User generated content • Video IP traffic will be 73% of all Internet traffic by 2017. The sum of all forms of video (TV, VoD, Internet, and P2P) ~ 80- 90% [Cisco, 2013] • 250+ billion photos uploaded and 350+ million photos uploaded every day [Facebook, 2013] • “Big data” is (mostly) visual data
  3. 3. How to exploit UGC So many applications But .. so much garbage
  4. 4. The human computation trust circle people data Passive crowdsourcing Active crowdsourcing Crowdsourcing optimization Incentives Trust computing Adversarial computing Provenance tracking Tampering detection Uncertainty modeling and reduction Semantic enrichment algorithms Reliability Optimization Predictive modeling Quality guarantee FOCUS HERE
  5. 5. Example: incentives • Problem: identify 10 balloons anchored in 10 undisclosed locations in the US, $ 40,000 prize to the winner • Solution in less than 9 hours • Recursive incentive mechanism (Nash equilibrium)
  6. 6. Another (different) incentive scheme • Complex content (3d with constraints) • Computationally intractable • Solved with Tetris-like game • Massive voluntary online collaboration, community quality monitoring
  7. 7. How to compute people Influence & Trust
  8. 8. How to fight adversaries Goal • Obtain quality content with minimum amount of human and computational resources • Algorithm can fail • But humans can cheat! Object detection example
  9. 9. In summary • Exploiting content requires .. good content • Computers and humans can cooperate in new ways – More than algorithm optimization – More than crowdsourcing • Old problem, but at a new scale – "On two occasions I have been asked, "Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?"... I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question." – Charles Babbage, Passages from the Life of a Philosopher (1864)