Scientists develop AI system capable of reading human emotions

A team of researchers led by professor Kim Hyun-bum from Incheon National University have developed a human emotion detection system using fifth-generation networks and artificial intelligence.

Researchers claim the virtual emotion detection system called 5G-I-VEmoSYS can recognize at least five kinds of emotion -- joy, pleasure, a neutral state, sadness and anger -- through body movements and other signals.

“Emotions are a critical characteristic of human beings and separate humans from machines, defining daily human activity. However, some emotions can also disrupt the normal functioning of a society and put people’s lives in danger, such as those of an unstable driver,” Kim said.

“Emotion detection technology thus has great potential for recognizing any disruptive emotion and in tandem with 5G and beyond-5G communication, warning others of potential dangers.”

So in the case of the unstable driver who could cause an accident, the car’s AI-enabled system can inform the nearest networks. Then nearby pedestrians would be notified of the potential danger via their smart devices.

The emotion detection system can also create a virtual emotion map by gathering a large amount of emotion data and this could be used to detect threats and prevent crimes.

The researchers said if the level of a serious emotion, such as anger or fear, became alarmingly high in a public area, the observed information could be conveyed to the nearest police station or relevant authorities who can take steps to handle risks.

The emotion detection system allows its process to be completed without revealing the face or other body parts of the human subjects to protect the privacy of individuals.

However, security issues about the emotion detection system remain, including the possibility of illegal signal tampering and the danger of sending false signals.

Although such concerns put the system’s reliability in question, the researchers are confident that they can solve the issues with further study.

“This is only an initial study. In the future, we need to achieve rigorous information integrity and accordingly devise robust AI-based algorithms that can detect compromised or malfunctioning devices and offer protection against potential system hacks,” Kim said.

The study was published in the journal of the Institute of Electrical and Electronics Engineers Network in October 2020.

By Kan Hyeong-woo (hwkan@heraldcorp.com)