USPTO Grants Patent for Augmented Neural Networks in Transfer Learning
Summary
The USPTO has granted a patent (US12585933B2) to International Business Machines Corporation for a method of transfer learning using augmented neural networks. The patent, effective March 24, 2026, details a technique for combining pretrained models with submodels trained on different domains to improve performance.
What changed
The United States Patent and Trademark Office (USPTO) has granted patent US12585933B2 to International Business Machines Corporation for an invention related to transfer learning with augmented neural networks. The patent describes a method where a pretrained model, trained on a first domain, is combined with a submodel trained on a second domain. This augmentation involves merging feature maps from both models, with the combined feature map then fed into a layer of the submodel. The patent was filed on September 5, 2019, and is effective March 24, 2026.
This patent grant signifies a new intellectual property asset for IBM in the field of AI and machine learning. While patent grants do not impose direct regulatory obligations on other entities, they establish exclusive rights for the patent holder. Companies operating in AI and machine learning, particularly those developing or utilizing transfer learning techniques, should be aware of this patent to avoid potential infringement claims. No specific compliance actions are required by this grant, but it highlights a patented technology that may influence future product development and licensing strategies in the AI sector.
Source document (simplified)
Transfer learning with augmented neural networks
Grant US12585933B2 Kind: B2 Mar 24, 2026
Assignee
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventors
Chungkuk Yoo, Bumsoo Kang, Minsik Cho
Abstract
A pretrained model is selected to operate in an augmented model configuration with a submodel. The submodel is trained using training data corresponding to a second domain, whereas the pretrained model is trained to operate on data of a first domain. The pretrained model is augmented, to form the augmented model configuration, with the submodel, by combining a first feature map being output from a layer in the pretrained model with a second feature map being output from a layer in the submodel. The combining forms a combined feature map. The combined feature map is input into a different layer in the submodel.
CPC Classifications
G06N 3/08 G06N 3/04 G06N 3/10 G06N 3/0454 G06N 3/082 G06N 20/00
Filing Date
2019-09-05
Application No.
16561896
Claims
20
Named provisions
Related changes
Source
Classification
Who this affects
Taxonomy
Browse Categories
Get Telecom & Technology alerts
Weekly digest. AI-summarized, no noise.
Free. Unsubscribe anytime.
Get alerts for this source
We'll email you when ChangeBridge: Patent Grants - AI & Computing (G06N) publishes new changes.