World

Scientists Create Artificial Neural Networks That Detect Human Ambiguity

Recently, artificial neural networks (ANNs) have become a powerful tool for researching artificial intelligence and data analysis. They are widely used in social sciences, robotics and neuroscience for classification, prediction and pattern recognition.
Sputnik

An international scientific team, which includes scientists from Russia’s Saratov State Technical University, has created an artificial neural network that detects human ambiguity.

In neuroscience, they help to classify neural signals, detect pathological activity of the brain (for example, with epilepsy), and neurodegenerative diseases. 

“However, despite significant achievements of artificial intelligence in recognition and classification of well-reproducible patterns of neural activity, the use of ANNs for recognition and classification of patterns in neural networks still requires additional attention, especially in ambiguous situations,” science publication CHAOS reported.

Why Neural Networks are Smarter Than Usual Computers
In this research, the scientists showed the efficiency of application of the ANN for categorizing human MEG trials corresponding to the sensitivity of bistable visual stimuli with different degrees of ambiguity. 

"In this way, we were able to recognize and characterize the state of uncertainty of human choice in detail at the time the decision was made by the brain," Head of the Scientific and Educational Center of Systems of Artificial Intelligence and Neurotechnology Professor Alexander Khramov said.

Moreover, the team was able to accurately measure the time period during which a person hesitates and cannot make a decision. 

According to the scientist, to identify such states of brain activity by traditional methods (for example, using frequency-time analysis) is quite difficult. But artificial intelligence proved to be very effective because it was able to show non-obvious signal features.

Who Dis? Smartphones Switch Off Our Brain, Make Us Do 'Stupid Things', Doc Warns
This data will help scientists create artificial intelligence (AI) which, in a complex situation will pause before making a decision. It will allow the AI to enter the waiting mode and accumulate information. 

“With the obtained results, we describe the possible application of ANNs for detection of bistable brain activity associated with difficulties in the decision-making process,” the publication read.

In practical terms this can be applicable in space missions. For example, a rover on Mars in a situation of doubt may seek help from operators on Earth. In this case, the artificial intellect of the rover must include a neural network which has a “stopping point” related to the impossibility to make a choice — that is, having doubts.

This can also be effective for testing and training the cognitive abilities of a person, his ability to make quick decisions in stressful situations. This is vital for many professions such as pilots, drivers, rescuers, military personnel or athletes. 

The authors of the study have already begun to develop appropriate training complexes based on these neuro-interfaces.

Discuss