Pretraining framework for neural networks
A processor has one or more circuits to use one or more neural networks to indicate an extent to which text corresponds to one or more images. The images may be medical images such as x-rays or CT scans, and the text may describe medical features such as medical anomalies or conditions. Features of...
Saved in:
Main Authors | , , , , |
---|---|
Format | Patent |
Language | English |
Published |
15.02.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | A processor has one or more circuits to use one or more neural networks to indicate an extent to which text corresponds to one or more images. The images may be medical images such as x-rays or CT scans, and the text may describe medical features such as medical anomalies or conditions. Features of the text and images may be calculated and processed using one or more multiplication and pooling processes, before being used to determine whether the text matches the image. The text and image features may be calculated using encoders. |
---|---|
Bibliography: | Application Number: GB20220009080 |