Learning & collaboration

 

 


     Collaborative filtering

      Can mutually mistrusting users collaborate in a distributed systems  ? The answer is yes. See following papers.

       2004 ACM Electronic Commerce: asynchronous trust system

 This paper introduces the notion that Byzantine participants cannot bring down a recommendation system. It shows that honest peers can still successfully cooperate to retrieve the best value.

       2005 IEEE ICDCS: concurrent trust system

       2005 ACM-SIAM SODA: collaborative filtering

       2005 ACM-SPAA Learning preferences by collaborating with strangers

This paper shows a randomized algorithm that successfully  reconstructs all values even in presence of (majority) of byzantine peers.

       2006 ACM-SPAA Collaborating with strangers in the presence of noise

        2007 ACM-SPAA Minimizing online regret against dynamic optimum

   


    Learning and adaptive systems

      Can one learn from past mistakes and predict an unpredictable  future  ? The answer is yes. See following papers.

       ACM STOC 1996: how to pick a winner

       ACM PODC 2003: learning best network path in presence of oblivious failures

       IEEE Infocom 2005:best network path in presence of adaptive failures,

       ACM STOC 2004: learning best network path with round-trip feedback

Some other useful papers and links


    Collaborative learning

      Can one combine learning with collaborative filtering  ? The answer is yes. See following papers.

       2005 ACM Computational Learning Theory: competitive collaborative learning

 


 

This material is based upon work supported by the National Science Foundation under Grant No. 0617883, 0515080,  0240551, 0311795

Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF).