| CPC G06F 3/04883 (2013.01) [G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06V 10/454 (2022.01); G06V 10/82 (2022.01); G06V 30/2264 (2022.01); G06V 30/2276 (2022.01); G06V 30/228 (2022.01); G06V 30/347 (2022.01)] | 20 Claims |

|
1. A computer-implemented method for training a neural network to generate a digital image of simulated handwriting, the method comprising:
receiving a handwriting sample image having a training sequence of characters;
receiving a coded text input also having the training sequence of characters;
processing the handwriting sample image using an encoder neural network to produce a style description vector that is representative of the handwriting sample image;
processing the coded text input using a decoder neural network to produce a raw generated image of the coded text input, wherein the decoder neural network uses the style description vector to produce the raw generated image;
evaluating a first difference between the handwriting sample image and the raw generated image; and
modifying the encoder neural network and the decoder neural network to reduce the first difference.
|