Taiwan high school student Chiu Hong-hsiang is winner of two awards at the I giovani e le scienze, or young people and science, competition for his Chinese font generation system, according to National Taiwan Science Education Center March 20.
Chiu, a senior at Taipei Municipal Jianguo High School, claimed gold in the prestigious event, as well as a special award from Yale Science and Engineering Association Inc. for his research during the annual competition staged March 16-18 in Milan.
Using the conditional generative adversarial network as a base, Chiu built a deep learning model for Chinese font style transfer. He also analyzed the effects of the U-Net—a convolutional neural network—and category embedding methods on the system.
According to Chiu, Chinese font generation is a time-consuming task and usually requires at least four to five years of work to design a new style for each of the 26,000 commonly used characters. The model quickens the process as it requires designers to develop a font for only a fraction of the characters, while the remainder can be transferred to the new font through the deep learning model, he said.
The NTSEC said Chiu’s accomplishment reflects the information and communication technology prowess of Taiwan’s young people, and is a vote of confidence in government efforts aimed at spurring creativity. (SFC-E)
Write to Taiwan Today at ttonline@mofa.gov.tw