Neural network capable of mimicking human handwriting presented
By ForGeeks Scientists from Muhammad bin Zayed University of Artificial Intelligence (MBZUAI) have developed a neural network that can create a font that looks very similar to human handwriting. They used several short pieces of handwritten text to train it.
In their study, the scientists used a neural network designed to solve computer vision problems. So far, they have focused on producing handwriting in English, but in the future they plan to train the algorithm in other languages, such as Arabic, which is difficult to analyze because of the way it is written.
The researchers used a neural network designed to solve computer vision problems.
The scientists compared the resulting method with other handwriting generation techniques. They showed the text generated by the three language models to a group of 100 people and asked them to choose the best option. In the end, 81% of the respondents preferred the new algorithm.
The developers believe this neural network will be useful for people who have lost the ability to write text by hand, as well as for generating large amounts of data to help improve the ability of machine learning models to process handwritten text. However, they are aware of the risks associated with the use of such a neural network by attackers, so no public launch date has been announced yet.
