/Patent Tracker
SamsungPatent applicationAI & softwareInterest 3/10

Samsung patents method to store sparse data matrices more efficiently

Samsung describes a technique for compressing large tables of data that contain many empty or zero values. The method stores only the non-zero numbers in a single list, then uses pointers and offsets to track where each number belongs in the original table.

Why it mattersThis is a routine software optimization that helps computers save memory and process large datasets faster, which is useful for background tasks in various devices.

In the filing’s own words

A method of storing a compressed matrix, including: identifying a plurality of non-zero data elements in an input matrix and generating a data array in which the plurality of non-zero data elements are arranged sequentially; generating a pointer corresponding to each row or each column of the input matrix; designating, to the pointer, a position in the input matrix or a number of non-zero data elements included in a corresponding row or column, wherein the pointer includes at least one of a row pointer and a column pointer; generating an offset indicating a relative interval between the plurality of non-zero data elements; and designating, to the offset, a relative index in a row or column direction, wherein the offset includes at least one of a row offset and a column offset; generating a compressed matrix; and storing the compressed matrix in at least one memory.

What exactly they claim (claim 1)

1. A method of storing a compressed matrix, the method comprising: obtaining an input matrix comprising a plurality of data elements; identifying a plurality of non-zero data elements in the input matrix and generating a data array in which the plurality of non-zero data elements are arranged sequentially; generating a pointer corresponding to each row from among a plurality of rows included in the input matrix, or corresponding to each column from among a plurality of columns included in the input matrix; designating, to the pointer, a position in the input matrix or a number of non-zero data elements included in a corresponding row from among the plurality of rows, or included in a corresponding column from among the plurality of columns, wherein the pointer comprises at least one of a row pointer and a column pointer; generating an offset that indicates a relative interval between the plurality of non-zero data elements, and designating, to the offset, a relative index for traversing the input matrix in a row direction or a column direction, wherein the offset comprises at least one of a row offset and a column offset; generating a compressed matrix corresponding to the input matrix based on the data array, the pointer, and the offset; and storing the compressed matrix in at least one memory.

Facts

Company
Samsung (SAMSUNG ELECTRONICS CO., LTD.)
Official title
MATRIX COMPRESSION AND TRAVERSAL METHOD AND ELECTRONIC DEVICE THEREOF
Publication number
US20260277784A1
Application number
19343777
Filed
Sep 29, 2025
Published
Sep 17, 2026
Inventors
1
Claims
20

Classified under

G06F 12/0207 Physics & computingG06F 2212/401 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Samsung 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.