Processing Parallelism for Machine Learning Model Training
Summary
The USPTO published patent application US20260093523A1 for a processing system that schedules parallel training of machine learning model instances based on microbatch counts. The system identifies expected idle cycles of processing units during training and schedules additional model instances during those periods. Inventors include Sumanth Gudaparthi, Yao Cui Fehlis, Karthik Ramu Sangaiah, and Sonali Singh. The application was filed on September 30, 2024.
What changed
USPTO published patent application US20260093523A1 for a method and system that optimizes machine learning training by scheduling parallel model instances during processing unit idle cycles. The system determines the number of microbatches required for training and calculates expected idle periods based on forward and backward pass times per microbatch. A scheduler then places second instances of the MLM into these identified idle cycles, improving computational efficiency. CPC classifications include G06F 9/4881, G06T 1/20, and G06N 20/00. Application number is 18901890.
Companies developing machine learning training systems should review this patent for prior art considerations in their R&D programs. Organizations implementing distributed or parallel ML training should assess whether their systems may practice claims covered by this application. While patent prosecution is ongoing, this publication establishes prior art as of the September 30, 2024 filing date. No regulatory compliance actions are required.
What to do next
- Review patent claims for overlap with current ML training infrastructure
- Assess prior art implications for in-house ML optimization projects
- Monitor application prosecution for granted claims that may affect licensing exposure
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.
PROCESSING PARALLELISM FOR MACHINE LEARNING MODEL TRAINING
Application US20260093523A1 Kind: A1 Apr 02, 2026
Inventors
Sumanth Gudaparthi, Yao Cui Fehlis, Karthik Ramu Sangaiah, Sonali Singh
Abstract
A processing system schedules parallel training of different instances of a machine learning model (MLM) based on a number of microbatches associated with training the machine learning model. The number of microbatches, along with the time required to complete a forward and backward pass of the MLM per microbatch, indicates the position, in time, of one or more expected idle cycles of a processing unit during training of a first instance of the MLM. A scheduler of the processing system schedules a second instance of the MLM during the one or more expected idle cycles.
CPC Classifications
G06F 9/4881 G06T 1/20 G06N 20/00
Filing Date
2024-09-30
Application No.
18901890
Named provisions
Related changes
Get daily alerts for ChangeBridge: Patent Apps - 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 plain-English summary, classification, and "what to do next" steps are AI-generated from the original text. Cite the source document, not the AI analysis.
Classification
Who this affects
Taxonomy
Browse Categories
Get alerts for this source
We'll email you when ChangeBridge: Patent Apps - AI & Computing (G06N) publishes new changes.