Neural network recogntion and training method and apparatus

This invention describes a method for training a neural network to recognize patterns. It works by feeding data into the network and then calculating a "loss" value that specifically encourages the network to correctly identify a target category while actively pushing it away from identifying a different, incorrect category. This unique loss calculation, which uses a probability that increases for the correct category and decreases for the incorrect one, is then used to adjust the network's internal settings, making it more accurate.

Why it matters: Filed before the widespread adoption of advanced contrastive learning techniques for deep neural networks. The specific loss function, which actively discourages an incorrect class while encouraging a correct one, aligns with principles that have since become crucial for training highly robust and performant models, particularly in self-supervised and representation learning.

Status: Not in forceListed as no longer active. The specific reason is not in the record we hold.
How hard to build: SpecializedNeural network software system

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