Neural Network Alignment Using Filter Layers
Summary
The USPTO published patent application US20260093990A1 describing methods for aligning pre-trained generative neural networks by introducing trainable filter layers. The filter layers process outputs from pre-trained network stacks while keeping the original pre-trained parameters fixed. Inventors: Xiangyu Qi, Xiao Ma, Ahmad Beirami.
What changed
The patent application discloses a method for aligning neural network outputs by adapting a pre-trained generative neural network through the introduction of one or more filter layers. Each filter layer processes outputs from stacks of pre-trained neural network layers and generates filtered outputs that are then fed to subsequent layers. The trainable parameters of the filter layers are adjusted using a training objective to increase aligned responses to training requests, while preserving the fixed pre-trained parameters of the base network.
This patent application does not impose any regulatory obligations or compliance requirements. Technology companies developing AI systems may consider this approach for aligning generative neural networks. The filing date was October 1, 2025, with publication on April 2, 2026. No immediate action is required by compliance teams.
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.
ALIGNMENT OF NEURAL NETWORKS USING ARCHITECTURAL MODIFICATIONS AND TRAINING EXAMPLES
Application US20260093990A1 Kind: A1 Apr 02, 2026
Inventors
Xiangyu Qi, Xiao Ma, Ahmad Beirami
Abstract
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium for aligning the output of a pre-trained generative neural network. In one aspect, the pre-trained generative neural network is adapted by introducing one or more filter layers. Each filter layer processes a filter layer input comprising an output from a stack of the pre-trained neural network layers, in accordance with trainable parameters of the filter layer, to generate a filter layer output. A next neural network layer after the stack of pre-trained neural network layers is configured to process at least the filter layer output. The trainable parameters of the filter layer(s) are adjusted using a training objective to increase the likelihood of the adapted neural network generating aligned responses to a plurality of training requests, whilst keeping pre-trained trainable parameters of the pre-trained neural network layers fixed.
CPC Classifications
G06N 3/082 G06N 3/084
Filing Date
2025-10-01
Application No.
19346875
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 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 USPTO Patent Applications - AI & Computing (G06N) publishes new changes.