Sign In

Communications of the ACM

ACM TechNews

Neural Networks Allow ­S to 'Read Faces' in a New Way

facial recognition, illustration

Neural networks have made technologies like facial analysis significantly more accurate in recent years.

Stanford University professor Michal Kosinski is using deep neural networks to illustrate the use of facial-recognition technology for invasive monitoring, among other things.

In one demonstration, Kosinski used facial analysis software called VGG-Face to extract data from approximately 35,000 headshot photos taken from a U.S. dating website, translate their attributes into a sequence of numbers, and then use a computer model to find relationships between sexuality and facial features. When presented with only one photo of each individual, the model differentiated between gay and straight men with 81-percent accuracy and with 74-percent accuracy for women, versus 61 percent and 54 percent when the task was performed by people.

Kosinksi says his latest challenge is to use similar software to determine subjects' political leanings. "I'm trying to tell the public that facial analysis technologies are [already] being used by companies and governments to invade privacy at an unprecedented scale," he says.

From Financial Times
View Full Article – May Require Subscription


Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found