US 12,079,699 B1
Verifying the provenance of a machine learning system
Jona Boeddinghaus, Staufen im Breisgau (DE); Mauritz Kop, Herten (NL); Paolo Campigotto, Vienna (AT); and Artur Susdorf, Breisach am Rhein (DE)
Assigned to DAIKI GmbH, (AT)
Filed by DAIKI GmbH, Vienna (AT)
Filed on Nov. 20, 2023, as Appl. No. 18/514,202.
Int. Cl. G06N 20/00 (2019.01)
CPC G06N 20/00 (2019.01) 20 Claims
OG exemplary drawing
 
1. A method of, comprising:
receiving a header associated with a machine learning system, metadata associated with the machine learning system, and content associated with the machine learning system;
generating a digital envelope corresponding to the machine learning system based on the received header, the received metadata, and the received content associated with the machine learning system, wherein the generated digital envelope corresponding to the machine learning system includes a first value corresponding to the header associated with the machine learning system and a combined value corresponding to the metadata associated with the machine learning system and the content associated with the machine learning system, wherein the first value corresponding to the header associated with the machine learning system is unencrypted and the combined value corresponding to the metadata associated with the machine learning system and the content associated with the machine learning system is encrypted;
storing the digital envelope corresponding to the machine learning system;
receiving a header associated with a second machine learning system, metadata associated with the second machine learning system, and content associated with the second machine learning system;
generating a second digital envelope corresponding to the second machine learning system based on the received header, the received metadata, and the received content associated with the second machine learning system, wherein the generated digital envelope corresponding to the second machine learning system includes an additional first value corresponding to the header associated with the second machine learning system and an additional combined value corresponding to the metadata associated with the second machine learning system and the content associated with the second machine learning system, wherein the additional first value corresponding to the header associated with the second machine learning system is unencrypted and the additional combined value corresponding to the metadata associated with the second machine learning system and the content associated with the second machine learning system is encrypted; and
comparing the first value corresponding to the header associated with the machine learning system to the additional first value corresponding to the header associated with the second machine learning system to determine a provenance and authenticity of the second machine learning system.