US Patent Application 17715798. Secure Multiparty Deep Learning via Shuffling and Offsetting simplified abstract
Contents
Secure Multiparty Deep Learning via Shuffling and Offsetting
Organization Name
Inventor(s)
Andre Xian Ming Chang of Bellevue WA (US)
Secure Multiparty Deep Learning via Shuffling and Offsetting - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 17715798 Titled 'Secure Multiparty Deep Learning via Shuffling and Offsetting'
Simplified Explanation
The abstract describes a method for protecting access to data samples when outsourcing deep learning computations. This is done by dividing each data sample into randomized parts and applying an offset operation to some of these parts. The modified parts are then shuffled with parts from other data samples and sent to external entities for deep learning computations. The order of applying the summation and deep learning computation can be changed. The results from the external entities are shuffled back, and the reverse offset and summation are applied to obtain the final result of the deep learning computation on the data sample.
Original Abstract Submitted
Protection of access to data samples in outsourcing deep learning computations via shuffling parts. For example, each data sample can be configured as the sum of a plurality of randomized parts. At least some of the randomized parts can be applied an offset operation to generate modified parts for outsourcing. Such parts from different data samples are shuffled and outsourced to one or more external entities to apply a deep learning computation. The deep learning computation is configured to allow change of the order between applying the summation and applying the deep learning computation. Thus, results of the external entities applying the deep learning computation to their received parts can be shuffled back for a data sample to apply reverse offset and summation. The summation provides the result of applying the deep learning computation to the data sample.
