Author : Chandan Kumar and Prof. Vandana Navale

Biometric distinguishing proof normally examines an enormous scale  database of biometric records for finding a nearby enough match of a person. This work researches how to redistribute this computationally costly checking while at the same time securing the privacy of both the database and the calculation. Abusing the intrinsic structures of biometric information and the properties of recognizable proof tasks, we first present a privacysaving biometric ID plot which utilizes a solitary server. We at that point think about its augmentations in the two-server model. It accomplishes a more elevated level of privacy than our singleserver arrangement expecting two servers are not plotting. Aside from to some degree homomorphic encryption, our subsequent plan utilizes clustered conventions for secure rearranging what’s more, least choice. Our trials on both manufactured and genuine datasets show that our answers beat existing plans while protecting privacy.

Keywords: Twitter, Location Inference, Bayes, LSTM