Algorithm can distinguish 'good general, generality'

The Chinese team developed an algorithm that is said to be capable of detecting the characteristics of the subject's face crimes at an almost 90% accuracy rate.

Two scientists from Shanghai Jiaotong University, China, have developed a new computer algorithm that is said to be able to identify criminals by analyzing their faces , IFL Science reported today. .

The top four types of face are common in crime, the bottom line is the three types of faces of law-abiding people. (Photo: Xiaolin Wu, Xi Zhang).

By examining the identity of 1,856 people, with half of them convicted, this algorithm can identify certain characteristics to distinguish crime generals from those who obey the law. French.

Based on factors such as the lip curvature, the angle of the inside of the eye, and the angle between the nose and mouth, algorithms can find the face of crime among thousands of photographs at nearly 90% accuracy.

According to the research team, the type of faces of criminals and ordinary people can be divided into two main groups. Law-abiding people have three types of facial structure, while the criminal group has four types.

"In other words, the faces of people in the law-abiding group have a higher degree of similarity than the criminal group , " the study's authors write.

This research has caused much controversy. Experts believe that this type of classification is completely unethical. It also has no scientific basis because all the relationships between physical characteristics and personality are denied.