Mar 09, 2021 17:00:00

Technology to read human brain waves and generate a favorite face is born

'Beautiful face' has been pursued and evaluated all over the world, such as the '

Top 100 Most Beautiful Faces in the World' being announced every year. However, people have different tastes, and which face is considered 'beautiful' is completely different depending on the individual sense. On March 5, 2021, Michel Speep and colleagues at the University of Helsinki announced that they had created a technique to read brain waves and create a person's favorite face.

Brain-computer interface for generating personally attractive images --IEEE Journals & Magazine
https://ieeexplore.ieee.org/document/9353984

Beauty is in the brain of the beholder: An AI generates personally attractive images by reading brain data | University of Helsinki
https://www2.helsinki.fi/en/news/data-science-news/beauty-is-in-the-brain-of-the-beholder-an-ai-generates-personally-attractive-images-by- reading-brain-data

Beauty is in the Brain: AI Reads Data, Generates Personally Attractive Images
https://www.laboratoryequipment.com/573909-Beauty-is-in-the-Brain-AI-Reads-Data-Generates-Personally-Attractive-Images/

Speep and colleagues first used a hostile generation network to generate hundreds of realistic facial photographs. A total of 30 subjects with electroencephalographs were shown several facial photographs and instructed each subject to pay more attention to the faces they found 'attractive'.

After that, the brain waves read were analyzed, and the characteristics that some faces whose active brain waves were measured from the subjects had in common were established, and a new face that combined these characteristics was created using AI.

Then, when the subject was shown the 'face that the person would like' generated for each subject at the same time as several other facial photographs, and asked to select one of the 'most attractive faces', about 87%. It is said that the face photo generated by AI was selected with the probability of.

'Although the experiment of generating images from objective elements was already successful, this experiment showed that images can be generated from subjective elements such as personal tastes,' said Speep. I have. Speep also hopes that the technology could help identify potential stereotypes and prejudices that individuals have.