Secure Artificial Neural Network Models in Outsourcing Deep Learning Computation: abstract simplified (17715835)

• This abstract for appeared for patent application number 17715835 Titled 'Secure Artificial Neural Network Models in Outsourcing Deep Learning Computation'

Simplified Explanation

The abstract discusses a method for protecting access to artificial neural network (ANN) models when outsourcing deep learning computations. This is achieved by dividing the ANN model into randomized parts, some of which are offset or encrypted. These modified parts are then shuffled and outsourced to external entities. Similarly, the data samples used as inputs to the ANN models are split into parts to protect them. The final result of a data sample applied to an ANN model is obtained by summing the responses of the model parts with the corresponding sample parts applied as inputs.

Original Abstract Submitted

Protection of access to artificial neural network (ANN) models in outsourcing deep learning computations via shuffling parts. For example, an ANN model can be configured as the sum of a plurality of randomized model parts. Some of the randomized parts can be applied an offset operation and/or encrypted to generate modified parts for outsourcing. Such model parts from different ANN models can be shuffled and outsourced to one or more external entities to obtain the responses of the model parts to inputs. Data samples as inputs to the ANN models can also be split into sample parts as inputs to model parts to protect the data samples. The result of a data sample as an input applied to an ANN model can be obtained from a sum of responses of model parts with the sample parts applied as inputs.