What actually is Machine Learning It's what we're going to see now. Here at Sandeco Channel Here we go! !
! Good Morning. .
. good afternoon. .
. Good night to you, you are now. watching another video from Sandeco Channel My name is Sanderson Macedo I am a professor and researcher in intelligence artificial and Data Science When I see the people talking about machine learning I see them explain machine learning is a subarea of the artificial intelligence that machine learning is a mix of computing probability and statistical And they rate Machine Learning in supervised learning unsupervised and by reinforcement Okay, this is all right.
but the most incredible conclusion I've heard about Machine Learning it was given by François Chollet François Chollet Ah . . .
I don't know, the name of this guy is French. Who he said that Machine Learning is a new paradigm of programming there I was like this: wow, man. It's very crazy.
It is as follows in programming in artificial intelligence by symbolic programming A versatile programmer develops programming rules that process data to produce useful user answers there he makes a series of "whiles" "If's" . . .
"For" that process data and return answers to the user already in Machine Learning There is a certain inversion in things the machine recive the same input data of classical formation but, instead of passing the programming rule as the classical formation the expected answers are passed then you ask: If I pass the answers What do we have as output? since in classical programming the data was passed plus the rules In Machine Learning the data is passed and answers And you know what comes out? The rules So these rules can be applied to new data to produce original answers and that's what Machine Learning is all about A machine learning system Is trained instead of being programmed explicitly Many relevant examples in a given assignment such as Pneumocad system.
I went through the various system images containing pneumonia and not pneumonia And I told the system which images had pneumonia and which images had no. There in the end the system learned "what pneumonia is" and when I send a new image to the system it's able to tell us if the image new image has pneumonia or not. Futhermore, if I pass more and more images to the system it will get better in this task.
Well, this is the formal definition Mitchell, right? a computer program learn from experience "E" the experiences here in this case are the X-ray images I passed to Pneumocad With respect to some type of task "T" The task here is to say if radiography has pneumonia or not if its performance "P" in the tasks in "T" performance is the percentage of Pneumocad hit the percentages of right to diagnose a new x-ray measured by "P" improve with the "E" experience this is exactly what I said if I pass more and more images to Pneumocad system it will learn more and more. Mitchell picked it up and made a definition more relax than he himself said which is as follows: "Machine learning is concerned with the question of how to build computer programs that improve automatically with more and more experiences then that's it Machine Learning has the task to create rules that were once made the versatile programmers and these rules allow full automation of a certain task such as automating tasks how to bookmark your vacation photos or predict how much you will sell next semester in your store or to know what the tomorrow's temperature or for example, predicts if an email is spam or not spam Somebody sais that the accumulation of loneliness It's neither receiving spam I hope you enjoyed this video.
