Neural Network Sparsity Methods for Computational Efficiency
Summary
USPTO published patent application US20260093967A1 for methods improving neural network efficiency through increased sparsity. The application, filed October 2, 2025, covers systems that use predictor values to determine item importance and selectively limit computation to a proper subset. Named inventors include researchers from Google and academic institutions.
What changed
USPTO published patent application US20260093967A1 titled 'Increasing Sparsity to Improve Neural Network Efficiency.' The application describes methods for storing parameter values including weight values and predictor values trained to predict importance levels of items processed by a neural network. The system generates outputs by determining values for items using predictors, selecting a proper subset based on threshold comparisons, and limiting computation to the selected subset.
This patent application publication does not impose any compliance requirements on regulated entities. Technology companies developing neural network systems may review the publication for competitive intelligence purposes. No action is required from compliance officers unless their organization has pending intellectual property interests in neural network optimization technologies.
Archived snapshot
Apr 2, 2026GovPing captured this document from the original source. If the source has since changed or been removed, this is the text as it existed at that time.
INCREASING SPARSITY TO IMPROVE NEURAL NETWORK EFFICIENCY
Application US20260093967A1 Kind: A1 Apr 02, 2026
Inventors
David Ethan Culler, Prateek Jain, Zhipeng Jia, Sanjiv Kumar, Jeremiah Willcock, Chong You, Shreya Pathak, Lin Chen, Xinnan Yu, Venkata Sesha Pavana Srinadh Bhojanapalli, Suvinay Subramanian, Felix Ren-Chyan Chern, Alek Alexandrov Andreev, Praneeth Kumar Netrapalli, Kan Wu, Henry Marc Levy
Abstract
Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for increasing sparsity to improve neural network efficiency. In some implementations, a system stores parameter values of parameter matrices of one or more layers of a neural network. The parameter values of the parameter matrices include (i) weight values of the one or more layers of the neural network, and (ii) predictor values that have been trained to predict levels of importance of items processed by the neural network. The system generates an output, including: determining a value for each of multiple items using the predictor values, selecting a proper subset of the items based on the values in the vector based on a threshold, and generating output of the one or more layers limiting computation based on the selected proper subset.
CPC Classifications
G06N 3/0499 G06N 3/048
Filing Date
2025-10-02
Application No.
19348252
Named provisions
Related changes
Get daily alerts for USPTO Patent Applications - AI & Computing (G06N)
Daily digest delivered to your inbox.
Free. Unsubscribe anytime.
Source
About this page
Every important government, regulator, and court update from around the world. One place. Real-time. Free. Our mission
Source document text, dates, docket IDs, and authority are extracted directly from USPTO.
The summary, classification, recommended actions, deadlines, and penalty information are AI-generated from the original text and may contain errors. Always verify against the source document.
Classification
Who this affects
Taxonomy
Browse Categories
Get alerts for this source
We'll email you when USPTO Patent Applications - AI & Computing (G06N) publishes new changes.
Subscribed!
Optional. Filters your digest to exactly the updates that matter to you.