Neural network output layer for machine learning

This describes a specialized part of an artificial intelligence system, specifically a neural network, that helps it make final decisions. It takes a list of raw scores from the network and converts them into a list of probabilities, where each probability is between zero and one, and all probabilities add up to one. The claims narrow this down to using a Softmax function, which is a common way to do this, and mention a "parabolic estimator function" for the underlying math.

Why it matters: Filed before the explosion of demand for efficient AI inference on specialized hardware. The focus on fixed-point calculations for a Softmax layer on reconfigurable fabric is now critical for deploying complex deep learning models, especially for edge AI applications where computational efficiency is paramount.

Status: Not in forceListed as no longer active.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, FPGA programming

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