US 12,088,726 B2
Systems and methods for predicting communication account identities across decentralized applications
Alex Reeve, San Francisco, CA (US); Harrison Dahme, Stateline, NV (US); Leo Chen, New York, NY (US); Akash Shah, Oakland, CA (US); Ming Jiang, Foster City, CA (US); Sid Shekhar, London (GB); and Zhicong Liang, Bellevue, WA (US)
Assigned to Coinbase, Inc., Oakland, CA (US)
Filed by Coinbase, Inc., Oakland, CA (US)
Filed on May 3, 2022, as Appl. No. 17/735,157.
Prior Publication US 2023/0362010 A1, Nov. 9, 2023
Int. Cl. H04L 9/32 (2006.01); G06N 20/20 (2019.01); H04L 9/00 (2022.01)
CPC H04L 9/3236 (2013.01) [G06N 20/20 (2019.01); H04L 9/50 (2022.05)] 20 Claims
OG exemplary drawing
 
1. A system for predicting communication account identities across decentralized applications based on aggregating traits of blockchain and off-chain operations using a multi-tiered machine learning architecture, the system comprising:
cloud-based storage circuitry configured to:
store a first tier of a machine learning architecture, wherein the first tier comprises:
a first machine learning model trained to predict communication account identity traits of a first type based on historic communication data of the first type;
a second machine learning model trained to predict communication account identity traits of a second type based on historic communication data of the second type; and
an oracle layer for the first machine learning model and the second machine learning model;
store a second tier of the machine learning architecture, wherein the second tier comprises a plurality of rule sets for predicting communication account identities;
cloud-based control circuitry configured to:
receive a first data feed, wherein the first data feed corresponds to a first type of communication data, wherein the first type of communication data is on-chain data, and wherein the first type of communication data indicates a blockchain network volume of pending communications;
receive a second data feed, wherein the second data feed corresponds to a second type of communication data, and wherein the second type of communication data indicates communication formats of pending communications;
generate a first feature input based on the first data feed;
generate a second feature input based on the second data feed;
input the first feature input into the first machine learning model to generate a first output, wherein the first output indicates a first communication account identity trait;
input the second feature input into the second machine learning model to generate a second output, wherein the second output indicates a second communication account identity trait;
generate, using the oracle layer for the first machine learning model and the second machine learning model, a third feature input based on the first output and the second output;
determine, based on the third feature input, a first rule set from a plurality of rule sets for predicting communication account identities;
receive a first communication;
predict a first communication account identity for the first communication based on the first rule set;
determine an identity probability at a first time based on the first communication account identity;
determine an off-chain operation required for the first communication based on the identity probability; and
determine a recommendation based on the off-chain operation required for the first communication; and
cloud-based input/output circuitry configured to generate for display, on a user interface, the recommendation based on the first communication account identity.