US 11,810,676 B2
Verified permissioned blockchains
Bishop C. Brock, Coupland, TX (US); Frank Liu, Austin, TX (US); and Thomas S. Hubregtsen, Hassfurt (DE)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Nov. 26, 2018, as Appl. No. 16/199,570.
Prior Publication US 2020/0167338 A1, May 28, 2020
Int. Cl. G06F 7/00 (2006.01); G16H 50/70 (2018.01); G06F 16/23 (2019.01); H04L 9/06 (2006.01); G06F 16/27 (2019.01); G06N 20/00 (2019.01); G06N 5/04 (2023.01); G06V 10/774 (2022.01); G06V 10/94 (2022.01); G16H 30/00 (2018.01); H04L 9/00 (2022.01)
CPC G16H 50/70 (2018.01) [G06F 16/2379 (2019.01); G06F 16/27 (2019.01); G06N 5/04 (2013.01); G06N 20/00 (2019.01); G06V 10/774 (2022.01); G06V 10/95 (2022.01); H04L 9/0637 (2013.01); G06V 2201/03 (2022.01); G16H 30/00 (2018.01); H04L 9/50 (2022.05)] 17 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving data for a machine learning model via a blockchain network, wherein the blockchain network enforce requirements that participants must provide a number of certified labelled examples in exchange for the number of inference requests;
distributing the data to a plurality of endorsing peers of the blockchain network wherein the endorsing peer verify client signature;
executing, by the plurality of endorsing peers, the machine learning model on the data to generate a plurality of inferences from the data via the plurality of endorsing peers of the blockchain network, respectively;
determining, by automated analysis, that a consensus does not exist amongst the plurality of inferences generated via the plurality of endorsing peers based on differences in the plurality of inferences generated by the plurality of endorsing peers and a policy that specifies a degree that the plurality of endorsing peers must agree to achieve the consensus;
labeling the data as unresolved when consensus does not reach;
collecting the inferences and committing transaction on the labeled data based on the different inferences of the plurality of endorsing peers to a blockchain ledger of the blockchain network; and
updating the ledger with new training example by the endorsing peers.