The ARI robot. created to help improve the care received by the elderly who live alone. BARCELONA CITY COUNCIL / Europa Press
If you like science fiction, you will have seen various omnipotent robots that act like people and, in fact, do any job much better than a being human. In 2001: an odyssey in space or in Blade Runner , different robots or totally intelligent systems appear, in leading roles, capable of making the most profitable decisions. In the cinema, robots are either sympathetic and dedicated to helping humans, or they are evil and powerful, the most dangerous enemies of the human race. These emotional qualities are very important in the development of modern social robots. To design artificial emotional intelligence , which helps robots to make decisions and improve their interaction with people, complex mathematical models are required .
In the age of artificial intelligence, little by little we are getting used to interacting with computers or robots in our daily life: in educational, therapy or rehabilitation and companionship contexts. However, they are not emotional robots. Research has neglected, for decades, the role of emotions, as it was considered that they did not allow scientific analysis. This has changed in recent years, after many psychological and cognitive science experiments revealed that emotions – feelings, moods, personality – are more than a form of expression of human beings: they have an important impact on the processes cognitive and decision-making of people. With this idea in mind, in 1995, Rosalind Wright Picard proposed a new research direction, affective computing , a computing that relates to, arises and influences emotions.
Since then, affective computing has studied and developed systems and devices for recognize, interpret, process and simulate human affection. In addition to providing the machines with emotions – which are expressed spontaneously – it also seeks to make them capable of interpreting the emotional state of humans and adapting their behavior to them, giving an appropriate response.
The more confident we are of the prediction, the more hopeful or fearful we feel. These two emotions can be treated as "expected", which are reflected through the expectation of the event in the forecast.
It is an interdisciplinary field in which mathematics, computer science, neuroscience, cognitive science and psychology converge. Mathematics – specifically, Bayesian inference and game theory – allow the development of models capable of coding emotional processes. Intuitively, we know that certain circumstances can induce a specific emotion. For example, hope precedes events that people long for; fear arises in the face of physical dangers or threats. Both feelings are experienced when we anticipate something good or bad in the future. The more confident we are of the prediction, the more hopeful or fearful we feel. These two emotions can be treated as "expected", which are reflected through the expectation of the event in the forecast. This model can be incorporated into intelligent machines, combined with machine learning algorithms capable of learning patterns of events, associations between objects and expectations.
In recent years we have focused mainly on developing mathematical models , based on risk analysis, focused to decision-making and its integration into programmable robots. The underlying decision-making model uses a multi-attribute expected utility function, with probabilities based on the adversary risk analysis framework. Given a history of interactions, ambient states and potential actions, the decision is anticipated through a joint distribution, with limited memory periods. We extend this decision principle to allow impulsive behavior in some specific cases, with intense emotions.
We still don't understand the complex feedback loops and decision-making that drive and control emotional responses.
But affective computing is still in its infancy. We still don't understand the complex feedback loops and decision making that drive and control emotional responses. One way of action could be to use the learning algorithms to train the simulation model of human behaviors. For this, it is necessary to improve the structure with precise measurements and with a comprehensive computational strategy.
Affective computers should not only better help humans, but also could have greater capacity to make decisions. These types of results will take the field of decision-making to a new level, allowing us to better understand human-robot interaction and producing new, right now, unexpected applications.
Si Liu is a postdoctoral researcher at the Shanghai University of Science and Technology
Café y Teoremas is a section dedicated to mathematics and the environment in which they are created, coordinated by the Institute of Mathematical Sciences (ICMAT), in which researchers and members of the center describe the latest advances in this discipline, share points of view encounter between mathematics and other social and cultural expressions and remember those who marked its development and knew how to transform coffee into theorems. The name evokes the definition of the Hungarian mathematician Alfred Rényi: "A mathematician is a machine that transforms coffee into theorems."
Editing and coordination: Ágata A. Timón García-Longoria (ICMAT)
You can follow MATERIA at Facebook , Twitter , Instagram or subscribe here to our
newsletter Source: Elpais