The neural network writes the names of dishes with Chinese calligraphy and abstract paintings

Over the past few years, computer scientists have been developing increasingly advanced and sophisticated

artificial intelligence tools thatcan solve a wide range of problems. For example, generative adversarial networks (GANs), machine learning models that learn to generate new data, including text, audio files or images. Some of these models can also be adapted for creative purposes, such as creating unique designs, songs or poems.

“In 2019, we collaborated with a restaurant inShanghai to showcase some artificial intelligence technologies to improve customer interaction and service quality. We then came up with the idea of ​​using artificial intelligence technology to create personalized abstract art based on the food customers order and present the artwork to entertain customers while they wait to be served."

Professor Harry Jiangnan Wang, one of the researchers who conducted the study

The original purpose of Wang's recent work and hiscolleagues had to associate the names of Chinese dishes with art. To do this, they decided to use Chinese calligraphy, as it is an invaluable part of China's cultural heritage, capable of expressing complex meanings in an aesthetic and artistic way. In addition, they drew inspiration from some of the paintings by abstract expressionist artists such as the American painter Franz Kline, executed in the 1940s and 1950s.

An example of a neural network

The researchers first compiled a dataset onChinese calligraphy, where each symbol has many different calligraphic images. They subsequently trained a conditional GAN ​​model from the dataset to create new Chinese calligraphy images that look like symbols in the training set but have never been encountered before.

This model, created by the researchers, computesthe similarity between the characters used to write the name of a Chinese dish and the characters in the new calligraphic dataset they compiled determines the most similar ones. Subsequently, the model creates a completely new character based on these similar characters.

“In our recent study, we were ableuse Chinese calligraphy as a special medium to connect words and images, and then used NLP and GAN technologies to create abstract art inspired by Chinese calligraphy. We are now planning to fine-tune our model to better control our results."

Professor Harry Jiangnan Wang, one of the researchers who conducted the study

The model also applies a range of aesthetictransformations - transferring a style based on famous paintings by artists such as Picasso, Rothko, Russo and Hockney, or transforming an oil painting and adjusting the ratio of white space.

When researchers tested their frameworkbased on GAN, they found that it can reliably create a unique and aesthetically pleasing art of Chinese calligraphy inspired by food and other words in Chinese, as well as introduce stylistic elements of Expressionist artists.

See also:

The Ministry of Health of Argentina disclosed data on side effects in those who received Sputnik V.

The platypus turned out to be a genetic mixture of mammals, birds and reptiles.

It became known why sperm move in a straight line.

Abortion and science: what will happen to the children who will give birth.