US 11,811,934 B2
Distributed machine learning via secure multi-party computation and ensemble learning
Ramtin Mehdizadeh Seraj, San Francisco, CA (US); and Nicholas Chow, San Francisco, CA (US)
Assigned to Dapper Labs, Inc., Vancouver (CA)
Filed by Dapper Labs Inc., Vancouver (CA)
Filed on Feb. 10, 2022, as Appl. No. 17/669,202.
Application 17/669,202 is a continuation of application No. 17/363,615, filed on Jun. 30, 2021, granted, now 11,265,166.
Claims priority of provisional application 63/046,362, filed on Jun. 30, 2020.
Prior Publication US 2022/0239487 A1, Jul. 28, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 21/64 (2013.01); H04L 29/06 (2006.01); H04L 9/32 (2006.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G06N 7/01 (2023.01)
CPC H04L 9/3218 (2013.01) [G06N 5/04 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
providing, by one or more processors of a smart contract platform, a smart contract to a plurality of modelers that each store corresponding private data to be used in training a corresponding predictive model included in the smart contract, each of the plurality of modelers training its corresponding predictive model in the smart contract based on the corresponding private data of that modeler;
receiving, by the one or more processors of the smart contract platform, input data from a requester device to which is to be provided a combined prediction to be generated from individual predictions to be outputted by the trained predictive models of the plurality of modelers;
distributing, by the one or more processors of the smart contract platform, portions of the input data among the trained predictive models by inputting a corresponding distributed portion of the input data into a corresponding each one of the trained predictive models in the smart contract, each one of the trained predictive models outputting a corresponding individual prediction based on its corresponding inputted portion of the input data;
generating, by the one or more processors of the smart contract platform, the combined prediction by combining the individual predictions outputted by the trained predictive models; and
providing, by the one or more processors of the smart contract platform, the generated combined prediction to the requester device.