USPTO Patent US12585947B2: Modifying Computational Graphs

ChangeBridge: Patent Grants - AI & Computing (G06N)

Published March 24th, 2026

Detected March 25th, 2026

Summary

The USPTO has granted patent US12585947B2 to Google LLC for methods of modifying computational graphs to efficiently handle communication between devices. This patent introduces send and receive nodes to subgraphs, shifting communication coordination away from the backend.

View original document View source feed page

What changed

USPTO patent grant US12585947B2, filed on August 3, 2021, and granted on March 24, 2026, details methods and systems for modifying computational graphs. The invention, assigned to Google LLC, involves inserting send and receive nodes into subgraphs to manage communication between unique devices performing different operations. This approach aims to improve efficiency by reducing backend coordination burdens, allowing for parallel processing while devices execute subgraphs.

This patent grant is primarily of informational value to technology companies and researchers in the AI and computing fields. It does not impose new regulatory obligations or compliance deadlines on regulated entities. However, it may inform future technological developments and intellectual property strategies within the industry.

Source document (simplified)

← USPTO Patent Grants

Modifying computational graphs

Grant US12585947B2 Kind: B2 Mar 24, 2026

Assignee

Google LLC

Inventors

Vijay Vasudevan, Jeffrey Adgate Dean, Sanjay Ghemawat

Abstract

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for modifying a computational graph to include send and receive nodes. Communication between unique devices performing operations of different subgraphs of the computational graph can be handled efficiently by inserting send and receive nodes into each subgraph. When executed, the operations that these send and receive nodes represent may enable pairs of unique devices to conduct communication with each other in a self-sufficient manner. This shifts the burden of coordinating communication away from the backend, which affords the system that processes this computational graph representation the opportunity to perform one or more other processes while devices are executing subgraphs.