US 12,306,875 B2
Multiple query projections for deep machine learning
Kapil Kumar, London (GB); Abhishek Majumdar, Santa Clara, CA (US); Nitish Aggarwal, Sunnyvale, CA (US); and Srimaruti Manoj Nimmagadda, Saratoga, CA (US)
Assigned to Roku, Inc., San Jose, CA (US)
Filed by Roku, Inc., San Jose, CA (US)
Filed on Dec. 22, 2023, as Appl. No. 18/394,103.
Claims priority of provisional application 63/459,536, filed on Apr. 14, 2023.
Prior Publication US 2024/0346082 A1, Oct. 17, 2024
Int. Cl. G06F 16/9032 (2019.01)
CPC G06F 16/90332 (2019.01) 22 Claims
OG exemplary drawing
 
1. A computer-implemented method for generating a prediction based on a query, comprising:
providing, by at least one computer processor, the query to a deep machine learning (ML) model; and
by the deep ML model,
generating a plurality of query projection embeddings by projecting the query into each of a plurality of different query embedding spaces, the projecting comprising applying a different hash function to the query for each of the plurality of different query embedding spaces, each different hash function being associated with a corresponding query embedding space of the plurality of different query embedding spaces; and
generating the prediction based at least on the plurality of query projection embeddings.