A few past talks

PacGAN: the power of two samples for generative adversarial networks

SIGMETRICS, Rising Star Award Lecture, June 2017 – Matrix Completion and Crowdsourcing

Breaking the bandwidth barrier: local adaptive entropy estimator

Achieving budget-optimality with adaptive schemes in crowdsourcing

  • Workshop on Statistical Physics, learning, inference, and networks, Les Houches, France, March 2017

  • ITA, UCSD, February 2017

  • Statistics Department, University of Wisconsin, February 2017 – Adaptive Crowdsourcing

  • CESG tele-seminar, Texas A&M, January 2017

  • NIPS workshop on crowdsourcing, Barcelona, Spain, December 2016 – Adaptive Crowdsourcing

  • INFORMS, Nashville, TN, November 2016 – Adaptive Crowdsourcing

  • Machine Learning seminar, USC, October 2016

Spy vs. spy: Rumor source obfuscation


Rumor source obfuscation
A novel approach to designing messaging protocols over social networks that hides the source of a message presented at WNCG seminar, UT Austin, September 2015

Differential privacy


Privacy region and its applications
Nexus of information and computation theories, Institut Henri Poincare, Paris, France, March 2016

  • ITA 2015 – the composition theorem

Rank Aggregation from Pairwise Comparisons

Designing Reliable and Efficient Crowdsourcing Systems


Iterative learning for reliable crowdsourcing systems
A novel approach to designing reliable and cost-efficient crowdsourcing systems presented at Twenty-fifth Conference on Neural Information Processing Systems(NIPS), 2011


Iterative learning from a crowd
Brief presentation of the crowdsourcing model and our learning algorithm at Interdisciplinary Workshop on Information and Decision in Social Networks(WIDS), 2011

Fast Singular Vector Computation for Extremely Large Matrices


Gossip PCA
Presentation at the ACM SIGMETRICS, 2011

Inferring Low-rank Matrices from Partial Information