Artificial Intelligence Has Learned To Predict Kisses and hugs

Researchers at MIT have developed artificial intelligence for analysing body language and predict human behaviour.
Established scientists looked neural network 600 hours of TV shows such as “Desperate Housewives” and “Office”. After that, the system introduced new video and was asked to predict the behaviour of the people involved in them. Artificial intelligence had to call the next event, such as a hug, a kiss or a handshake.

0

5% Discount Coupon: Unknown Tentsile

Artificial Intelligence Prediction Accuracy

Neural Network, held a profound training anticipate events correctly rollers in 43% of cases. Of course, it is far from perfect, because living people were more effective predictors. People were able to correctly identify the next event in 71% of cases.


Another Photo from movie Ex Machina

WE WANTED TO SHOW THAT BY SIMPLY WATCHING TV SERIES, A COMPUTER CAN GET ENOUGH KNOWLEDGE TO EFFECTIVELY PREDICT HUMAN BEHAVIOUR

– said Carl Vondrik, a graduate student in computer science and artificial intelligence.

People use their own experience to predict the behaviour of others. But computers practically have no possibility of direct contact with the physical world.

Check: New study claims, You will be ‘romancing’ robots in 2025

Despite the fact that by Vondrikom and algorithms team is not efficient enough for practical use, development is another impressive illustration of the possibilities of artificial intelligence.

Perhaps someday artificial intelligence will create robots for emergency situations. Those machines will analyse the behaviour of a person and determine his psychological state.

TO ACCURATELY UNDERSTAND AND PREDICT HUMAN INTERACTIONS NEEDED MORE WORK. IN THE NEAR FUTURE WE WILL TEACH THE SYSTEM TO ASSESS THE SITUATION MORE SUBTLY

Vondrik said.

Companies such as Google, Facebook, and the Microsoft, are also working on the creation of artificial intelligence over the past few years. Due to their design users, in particular, got the tools for automatic classification and systematisation of personal photos. Via Technotification