[MUSIC] Hi, welcome again. We are starting today the last set of sessions of this course in which we'll try to understand what is happening during the 21st century on the study of emotions and how these studies are really being implemented into computers and robots. Today, we will talk about Neural Nets. When the beginning of the 20th century, John von Neumann created a metaphor of the brain as a computer It is that in a true life that was really useful stated that notion of the processing information metaphor between the relationship of some basic machine mechanisms that were able to create meaning of the world. So when Started the discipline of the artificial intelligence. Several others tried to make smart machines. And there were created several kind of strategies. In order to achieve this. One of them were the neural nets. We will see that there is a relationship between what the neurologists were obtaining about the functioning, functional properties of the brain and how these ideas were translated, for example, into neural networks with drive nodes, for example, in 1972 by Stephen Grossberg Or even when Dimassio created this new product label for search in 1994, very close to him. Rosalind Picard at MIT created the domain of Affective Computing that was also influencing research of another MIT researcher called Cynthia Breazeal that made social robotics. And even during the 80s of the 20th century, people like Ekman, Ortony, Scherer, or Oatley, among others, creating a new view on the role of emotions into cognitive processes, and it created the necessity of implementing emotions into emotional cognitive architectures. So, when we are talking about neuron links, we are talking about computation of tools inspired by a neural functional of human brains. And, again, they are trying to make systems of deficient systems who are able to learn by themselves with supervisor and supervised in our several strategies, but artificial systems are unable to learn things about information that it's arriving to these layer of processing units. All this started in 1943 when McCulloch and Pitts created Neural Networks. Later a psychologist called Donald Hebb fixed the possible mechanisms of learning based on neural plasticity that existed in our brains, and make possible new ways of creating neural networks able to learn. This all was even highly criticized by Minsky and Seymore in 1969, in which they try to understand that neural link's had some kind of really difficult ways of understanding some kind of tasks. But later, thanks to new techniques and new computational power, was able to create, for example, things to connect in here, new ways of understanding these neural nets so in these moments, neural networks are used two very important things like deep learning. So, today we're going to see a lot of neural networks, architectural, and the different neural network ways of understanding these kind of techniques. And you can see here that there are several architectures that have some, more or less, biological expression, and they have more good or not good result, et cetera, perform object activities testings. But, anyhow, the point is that we have emotional neural networks. Neural networks really deeply inspire on emotional, functional properties inside the brain. We can find several books following this kind of ideas. For example, Emotional Intelligence of Chakraborty and Kenar, or, People who is talking on emotional cognitive neural algorithms and with really dire engineering applications. You can see here two examples, two books that are working on this new perspective and, about how new logic, about new ways of processing information inspired by emotional mechanisms can really improve the design of new machines. So, neural nets were initially brain and emotionally inspired, and neural nets today are very useful for AI systems. And finally, near-emotional neural networks are really better than classic ones. And so from a functional perspective they are performing much better classic tests than other ones. So, surely emotional Implementation of emotional mechanisms inside these strategies and these architectures is really a good tip on how to do that. So thank you so much and hope to see you in the next video. Bye.
