Privacy-Preserving Belief Propagation and Sampling
Michael Kearns, Jinsong Tan, and Jennifer Wortman University of Pennsylvania (Poster T46) Why do we need privacy?
Do I have the disease?

· Our solutions fuse tools from learning and cryptography ­ Inference in belief networks ­ Secure function computation · The result is a distributed protocol that remains faithful to the network topology · Need more details? ­ Call 1-800-MY-PRIVACY ­ Or visit our POSTER!!!

Is my friend infected? How about his friend?

· What is the probability p that I have been infected? · Can my neighbors help me calculate p without revealing any other information?