Microsoft Technology Licensing, LLC

Ilya Mironov

Ramarathnam R Venkatesan

Eliot Carleton Gillum

Frank David McSherry

Anton Mityagin

Aditya Bansod

Yaacov Nissim Kobliner

Chun Yu Wong

User rating data may be received at a correlation engine through a network. The user rating data may include ratings generated by a plurality of users for a plurality of items. Correlation data may be generated from the received user rating data by the correlation engine. The correlation data may identify correlations between the items based on the user generated ratings. Noise may be generated by the correlation engine, and the generated noise may be added to the generated correlation data by the correlation engine to provide differential privacy protection to the user rating data.

Differential privacy preserving recommendation

An addition chain is first generated, and then an integer x is derived from it. Doubling and star steps may be implemented in the addition chain. This approach eliminates the computationally expensive step of generating the addition chain from an exponent, and therefore can greatly reduce the computation time of the modular exponentiation.

Systems and methods for generating random addition chains

A system and method for verifying messages. The method may include the steps of receiving an inbound message and characterizing the inbound message by analyzing a latent cryptographic identifier in the inbound message. The identifier is generated by a recognized message system, which may be the receiving system itself, for an outbound message. Characterizing may involve detecting if the latent cryptographic identifier is present and determining if the cryptographic identifier is valid. The step of determining can be performed using symmetric or asymmetric methods of verifying the authenticity of the message.

Validating inbound messages

Techniques are disclosed to limit short-term correlations associated with outputs of stream cipher keystream generators. Output values of a generator are paired such that the paired outputs are sufficiently far apart to be considered independent. In one described implementation, a method includes sequentially storing a plurality of results provided by a stream cipher output rule in a first, second, and third storage units. A pairing function pairs individual values from the first and third storage units that are at least a threshold value apart. Upon reaching the threshold value of the output rule results, the contents of the first, second, and third storage units are rotated serially.

Growing community of inventors

Mountain View, CA, United States of America

Ilya Mironov