Feature analysis in computer-based reasoning models

This invention describes a way to make artificial intelligence (AI) models more effective and balanced. It works by having a computer system analyze the different pieces of information, or 'features,' that an AI model uses. Based on how 'convinced' the model is by each feature, or how much it contributes, the system decides whether to keep or remove that feature, leading to a more refined and efficient AI. The claims specifically focus on applying this technique to systems that label images or control vehicles.

Why it matters: Since 2018, the complexity and scale of AI models, especially in autonomous systems and image recognition, have exploded. This invention's method for efficiently selecting and balancing features is now crucial for managing performance, resources, and ensuring robust AI behavior in these advanced applications.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedComplex software logic, machine learning expertise

3 AI reinvention directions for this patent

See all three

AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.

Reinvent this with AI

Where this one came from

Every morning we go through millions of abandoned patents and pick out 15 worth your time. This was one of today’s.

This one

+10