US Patent Issued to The United States of America As Represented By The Director Of The National Geospatial-Intelligence Agency on Jan. 13 for "Machine learning process to automatically label and identify mislabeled documents at large-scale" (Maryland Inventor)

ALEXANDRIA, Va., Jan. 13 -- United States Patent no. 12,524,664, issued on Jan. 13, was assigned to The United States of America As Represented By The Director Of The National Geospatial-Intelligence Agency (Springfield, Va.).

"Machine learning process to automatically label and identify mislabeled documents at large-scale" was invented by Garret Vo (Gaithersburg, Md.).

According to the abstract* released by the U.S. Patent & Trademark Office: "Systems and methods for automatically labeling documents and identifying mis-labeled documents at scale. Documents are pre-processed, divided into training and testing sets, and processed. After processing, a network graph is formed, the documents are embedded using a graph convolutional neural net...