US 12,476,787 B2
Homomorphic encryption
Henry Markram, Lausanne (CH); Felix Schuermann, Grens (CH); Kathryn Hess Bellwald, Aigle (CH); and Fabien Delalondre, Geneva (CH)
Assigned to INAIT SA, Lausanne (CH)
Filed by INAIT SA, Lausanne (CH)
Filed on Apr. 5, 2023, as Appl. No. 18/295,959.
Application 18/295,959 is a continuation of application No. 16/356,391, filed on Mar. 18, 2019, granted, now 11,652,603.
Prior Publication US 2023/0370244 A1, Nov. 16, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. H04L 9/00 (2022.01); G06F 17/18 (2006.01); G06F 21/60 (2013.01); G06N 3/047 (2023.01)
CPC H04L 9/008 (2013.01) [G06F 17/18 (2013.01); G06F 21/602 (2013.01); G06N 3/047 (2023.01)] 19 Claims
OG exemplary drawing
 
1. A method implemented in hardware, in software, or in a combination thereof, the method comprising
homomorphically encrypting secure data, comprising
determining whether signal transmission activity between nodes in a recurrent artificial neural network comports with pre-defined patterns of signal transmission activity between pre-defined groups of nodes, wherein the signal transmission activity is responsive to input of the secure data into the recurrent artificial neural network and signal transmission activity between a pre-defined group of nodes in the recurrent artificial neural network comports with a respective pattern of the pre-defined patterns of signal transmission activity when the signal transmission activity between the pre-defined group of nodes matches the respective pattern, and
storing binary data as a homomorphic encryption of the secure data, the binary data comprising a vector of digits comprising ones and zeros, wherein
each element of the vector corresponds to a respective pattern of the pre-defined patterns of signal transmission activity for a respective pre-defined group of nodes,
each non-zero element of the vector of digits in the binary data indicates that the signal transmission activity in the recurrent artificial neural network comports with the corresponding respective pattern of the pre-defined patterns of signal transmission activity for the respective pre-defined group of nodes in the recurrent artificial neural network; and
making the stored binary data available for statistical analysis that draws conclusions about the secure data.