Artificial Intelligence develops a neural network that can separate criminals and non-criminals by their mugshots.
It might sound like a scene from Minority Report, but it is not. Scientists from Shangai Jiao Tong University identified offenders with an accuracy of 89.5 percent via machine-vision algorithms. The study named ‘Automated Inference on Criminality’ is the first automated work that takes account of criminality in relation to still pictures of faces.
Artificial intelligence learns the common facial features of criminals
Criminologists integrate latest technologies to collect detailed data to identify criminals. According to Xiaolin Wu and Xi Zhang, scientists who ran the study, their method is straightforward. They first took ID photos of the criminals and non-criminals half and half. The mixture included 1856 Chinese men. These men were all in between 18 to 55 years old and without facial hair. Scientists used 90 percent of the photos to create a convolutional neural network. And the rest 10 percent was used to test the efficiency of the informed system.
Convolutional neural network identifies criminals correctly with 89.5 percent accuracy
The results were unsettling. Xiaolin Wu and Xi Zhang found that the neural network they created can identify criminals correctly with a 89.5 percent accuracy. Furthermore, they discovered a few morphologic features that are discriminative for predicting criminality. These features include inner corner distance of the eyes, lip curvatures, and nose-mouth angle. The study concludes:
“Above all, the most important discovery of this research is that criminal and non-criminal face images populate two quite distinctive manifolds. The variation among criminal faces is significantly greater than that of the non-criminal faces. The two manifolds consisting of criminal and non-criminal faces appear to be concentric, with the non-criminal manifold lying in the kernel with a smaller span, exhibiting a law of normality for faces of non-criminals.”
[Image Source: Pixabay]
Characteristics of a criminals face
According to the study, neural network recognized these on a criminal’s face:
The distance between inner corners of the eyes is 6 percent shorter.
The curvature of the upper lip is about 23 percent larger.
The angle between two lines drawn from the corners of the mouth to the tip of the nose is 20 percent smaller.
“We are the first to study automated face-induced inference on criminality free of any biases of subjective judgments of human observers. By extensive experiments and vigorous cross validations, we have demonstrated that via supervised machine learning, data-driven face classifiers are able to make reliable inference on criminality. Furthermore, we have discovered that a law of normality for faces of non-criminals. After controlled for race, gender, and age, the general law-abiding public has facial appearances that vary in a significantly lesser degree than criminals.”
The use of Artificial Intelligence, of course, brings ethical controversies with it. And raises questions about what is normal and what is not.
To get more information about the study visit Automated Inference on Criminality.