What if your computer or smart device could look you in the face, dig through your text messages and phone calls, review your purchasing patterns and determine your state of mind?

Some computers already are predicting users’ emotional states, thanks to advances in technologies such as facial and vocal recognition, artificial intelligence and data science. Though these projects are still in their infancy, potentially great payoffs loom in areas ranging from mental health to market research. But there are questions: Given the personal nature of the data, how will these technologies be used and how will people respond to them?

In the more promising applications, technologists, researchers and health care providers are hopeful that technology will be able to predict the emotions of users in situations where humans can’t be present — serving as an early warning system for people in need and perhaps even saving lives.

Learning emotions

Teaching computers to recognize human emotion is no small feat, but projects such as ICT’s and those underway by Waltham, Massachusetts-based Affectiva are making progress.

Affectiva has been able to train computers to recognize facial expressions, co-founder and chief science officer Rana el Kaliouby said. By scanning and storing enough examples of faces from people of varying ages, ethnicities and genders, a computer can identify individual features, classify faces according to common expression types and distinguish one expression from another, she said. Probability and machine learning are used to predict the emotion.

“[The computer] classifies it as a smile example with a certain probability, so if it’s really confident that it’s a smile, it will be 90 percent confident,” el Kaliouby said. “If it’s like, ‘Hmmm, this is a really subtle smile or the video’s too dark or if it’s not quite sure what the person’s doing,’ the probability might be 10 percent.”

By building a global-scale expression repository, Affectiva believes it ultimately will be able to use facial recognition to interpret and track people’s emotional states, with applications in such areas as conducting market research or treating mental illnesses.

Full Article