What is Machine Learning and Deep Learning?

If we are going to start a trip to the bowels of the field of artificial intelligence, we must first know the terrain through which we are moving and for that we need a map I do not mean a physical map but a conceptual map to have clear in our head exactly to what we mean when we talk about artificial intelligence machine learning big data neural networks or deep learning many concepts that normally overlap or are interpreted differently and that in the end end up being used with much confusion actually try to give an exact definition of what is artificial intelligence is a very complicated task especially because it is a concept that depends on the definition of intelligence itself that still has multiple interpretations for this reason it happens that when we want to try to define what artificial intelligence is we also find many authors who define it in their own way if we take all these d efinitions and we draw a common idea we have that artificial intelligence is the discipline of the field of computer science that seeks to create machines that can mimic intelligent behaviors, these behaviors can be very diverse lead to analyze patterns recognize voices or win games are many ways in which a machine can simulate intelligent behavior and we have more and more examples of how in certain areas they manage to achieve greater performance to the human but that makes them more capable than us well not exactly take any of these guides that stand out in a very specific domain and try to perform another task you will see that the result you get is disastrous this ability to be able to perform multiple tasks is what allows us at the same time to think about walking and talking when for example we are chatting on the street with a friend and it is a very coveted feature that is still inv Stressing in all the departments of artificial intelligence this leads us to a first classification of the types of weak and strong aunts we speak with artificial intelligence is weak to refer to those systems that can only fulfill a very limited set of tasks for example as much as you have taught your robot to walk how you try to do something different like shutter a ball you may not get the expected result on the contrary the strong ones refer to those aunts who are able to apply to a variety of different problems and domains you have to know that today all our guides are still classified in the first group, no matter how much Hollywood has reached the second one, many years ago we go back to the definition of before it is important to emphasize the imitation of intelligent behaviors because it is the key to understanding the other concepts that we are going to try to notice that imitating does not mean that That this behavior is essentially cognitive behavior, that is to say.

 I can program the movements of a robotic arm in a classical way so that they always carry out the same movement that in principle does not seem very intelligent since the logic of the movement of the programmed I but nevertheless This fits within the definition that we have given since apparently the machine performs intelligent behavior seen in the field of artificial intelligence, we can find different subcategories that respond to different intelligent behaviors, for example, as we just mentioned if we look at the capacity to move and adapt to the environment we deal with the field of robotics or if we refer to the ability to understand language we have the field of natural language production or the ability to speak a field in which both voice conversion is studied text or text-to-speech conversion all these capabilities confo They have their own fields of study in the world of artificial intelligence but nevertheless if there is a capacity that really defines us as intelligent agents, it is the ability to learn, that is, machine learning, match in learning or machine learning is the branch of intelligence. 

Artificial that studies how to provide machines with a learning capacity understood as the generalization of knowledge from a set of experiences, this learning, as explained in a previous video, can be divided into three different groups supervised learning, not supervised and reinforced byWe already know that margin learning is a discipline within the field of artificial intelligence, but it is not just any discipline, it is a nuclear component that in fact relates to the rest of the categories, because obviously all other capabilities can be limited either because someone has programmed them or much more interesting because the system itself has learned to perform the wolves by catching one thing is to program a machine so that it can move and very different ones to program it so that it learns to move equally it is not the same to program what elements make up a face that automatically learn what is a face this paradigm shift what makes learning the most interesting and therefore it is very common to confuse the part for the whole and say that artificial intelligence and machine learning is the same thing luckily this you know the difference within the match in learning we find a new world where there are Different techniques that serve to cover different types of applications, for example, we have techniques such as decision trees, regression models, classification models, clustering techniques and many others, however, if one of these techniques has given fame to the field of may in learning during the In the last decade this has been the neural networks. The interesting thing about neural networks is that they are able to learn in a hierarchical way, that is, information is learned by levels where the first layers learn very specific concepts such as a screw, a mirror, a wheel. in the later layers the previously learned information is used to learn more abstract concepts such as a car a truck a motorcycle this makes that as we add more layers the information that is learned is increasingly abstract and interesting and how many layers can we put then there is no limit and the tendency is that each time these algorithms we add more and more layers becoming increasingly complex algorithms this increase in the number of layers and complexity is what makes these algorithms known as deep learning algorithms sure that this dealer no or deep learning I have already heard before and it is a concept that has become very fashionable in recent years but we really need these complex techniques because the reality is that if you think that these techniques are trained and learned from the data and we are currently immersed in the information age with the advent of digitalization, the reduction of storage devices and a change in mentality when appreciating the value of data, we have entered into a tendency to accumulate more and more data, what has been called big data therefore and to summarize we have that big data refers to the phenomenon of accumulating large amounts of data but also It is usually used to refer to the process of analysis of all this data from the moment in which two captures until you transform them into knowledge for this, they require powerful and complex techniques such as deep learning techniques, these are only a version Vitamins of neural networks a family of machine learning algorithms that have given a new resurgence to the field of machine learning and therefore to the field of artificial intelligence.