/Patent Tracker
GoogleGranted patentAI & softwareInterest 4/10

Google patents method to train AI models directly inside data warehouses

This patent describes a system that converts text data into numerical formats, known as embeddings, without moving the data out of a central storage system. It then uses these numbers to train machine learning models, allowing the entire process to happen within the existing data infrastructure.

Why it mattersThis approach could make it faster and cheaper for companies to build AI tools by avoiding the slow and expensive step of moving large datasets between different servers.

In the filing’s own words

A method includes receiving a text embedding generation query from a user requesting generation of a text embedding for one or more data elements stored at a data warehouse. In response, the method includes selecting, using the text embedding generation query, a text embedding model from a plurality of different text embedding models. The method includes generating, using the selected text embedding model, the text embedding for the one or more data elements and storing the text embeddings at the data warehouse. The method includes receiving a machine learning model training query from the user device requesting training of a machine learning model using the text embeddings. In response to receiving the machine learning model training query, the method includes training the machine learning model using the text embeddings. The method includes providing, to the user device, a notification indicating that training of the machine learning model is complete.

What exactly they claim (claim 1)

1. A method comprising: receiving, by data processing hardware and from a client device, a text embedding generation query requesting the data processing hardware: generate, based on a computational cost threshold, a text embedding for each data element of a plurality of data elements stored at a data warehouse; and train a machine learning model using the text embeddings; responsive to receiving the text embedding generation query, selecting, by the data processing hardware and using the computational cost threshold, a text embedding model from a plurality of different text embedding models integrated in the data warehouse; generating, by the data processing hardware and using the text embedding model, the text embedding for each data element of the plurality of data elements, wherein the text embedding model executes within the data warehouse without migrating the plurality of data elements out of the data warehouse; training, by the data processing hardware and using the text embeddings, the machine learning model; and after training the machine learning model, storing, by the data processing hardware, the machine learning model at the data warehouse.

Facts

Company
Google (Google LLC)
Official title
Text embedding generation and applications
Patent number
US12743417B2
Application number
18484661
Filed
Oct 11, 2023
Granted
Sep 22, 2026
Inventors
6
Claims
20

Classified under

G06F 16/24522 Physics & computingG06F 16/2433 Physics & computingG06F 40/40 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Google in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.