US Patent Application 17715885. Non-uniform Splitting of a Tensor in Shuffled Secure Multiparty Computation simplified abstract
Contents
Non-uniform Splitting of a Tensor in Shuffled Secure Multiparty Computation
Organization Name
Inventor(s)
Andre Xian Ming Chang of Bellevue WA (US)
Non-uniform Splitting of a Tensor in Shuffled Secure Multiparty Computation - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 17715885 Titled 'Non-uniform Splitting of a Tensor in Shuffled Secure Multiparty Computation'
Simplified Explanation
The abstract describes a method for protecting access to the values of elements in a tensor during the outsourcing of deep learning computations. The tensor is divided into portions, and some of these portions are further split into parts. These parts are used to generate computing tasks, where each task operates on a portion or part of the tensor. Some portions may share common parts. The computing tasks are generated based on unique parts to avoid redundant computations. The tasks are shuffled and distributed out of order to external entities. The final result of operating on the tensor is obtained by combining the results received back from these external entities.
Original Abstract Submitted
Protection of access to values of elements in a tensor in outsourcing deep learning computations. For example, the tensor in the computation of an artificial neural network can be partitioned into portions. Some of the portions can be selected for splitting into parts, such that the sum of a set of parts is equal to a respective portion being split to generate computing tasks. Each computing task is configured to operate based on a portion of the tensor or a part of a portion of the tensor. Some of the portions may share common parts. The computing tasks can be generated according to unique parts to eliminate duplicative computing efforts. The computing tasks can be shuffled for distribution out of order to external entities. The result to operate on the tensor can be obtained from results, received back from the external entities, of the outsourced computing tasks.
