I’m first taken through the history behind this big new leaf book What exactly happened One of the limitations of this very early prototypes which they came out with on how did it all evolve And today why is it such a popular Why’s it so popular that everything that you see today you Racine Alexa you’re a sin city Us in Google assistant Last week I was in China on board The writing was in Tamil I couldn’t understand anything There was this guy foreigner He just pointed this theme of elected and he went away He knew what the message was And I’m still waiting for the musical that is online Real time language translators.
You point a phone towards it Translated into English All these technologies are based on this algorithm called They’re variants off deepness Littles Of course we have such for example are in and recommend new networks We have something called LSD in long short term memory So down the line you’ll come across all these things Very interesting field but it in just a lot of power popularity But the power can be a boon And being both you miss applied it’ll completely mislead you So before you build your models and trying to put these models in the production here to do a lot of testing.
Welcome to the small of Bittner Lynbrook Before I start on this any questions you have from anything that we’re done below just in case because all these things are connected to each other Heaven linear regression You’ll see linear regression coming to play heroes You’re gonna logistical regulation all the you’ll see logistic also come into play in this not directly indirectly logistic curve you’re seeing logistical It’s also called Cigarette Girl You know the properties of signaling goof What is important range What ISO could reach seen manage function manages cousin off signore it It also looks the same in port is minus infinity plus infinity output.
This minus one plus one Yep these two functions played a very important role in the history of deep personal effects But today we rarely use these to function than this because of certain limitations of these two functions Yep So we’ll understand all those things and we’ll start No But before I start once again any question shoot Otherwise we’ll great No it is It is an advancement because along with support vector machine this one is the most powerful algorithm available to us in danger since truly still a few years ago If you if this was 2010 for example 2010 not very far back newly troupe was not very popular The reason for that is we didn’t have the hardware technology with us Tobe able to implement New really works It is only after cloud computing.
You’re big data technologies on your Internet became robust enough All these technologies when they converged did we find the required context in which we’ve woken you neural networks The results started in 19 for peace so it’s not something new It’s actually started in 19 forties but during those times they did not have the kind off support required in 19 forties And your little went through a lot of up sandals until 10 4012 where it really started picking up on it left all the algorithms behind.
What you’re a little can do even support vector machine cannot No that is also very powerful So this stands on the top followed vestry um in terms off the power to do things followed the other end Well they’re very relevant A ce we speak now They’re lost being formulated in far East on in Europe where the law requires that when you do modeling your model should be white box it means you should be able to explain on the world conditions You took this decision whether the person is going to turn or not before before Nunally Book is a black box.
It’ll tell you the result But it looked really why it reads to the country’s You’re not We will find the book so other undertones of quite relevant.
If you’re working in finance banking insurance well most of the time find the other vehicles especially decision trees It’s barely write books right in execution like any other eligible This algorithm is also part of surprise Ling returned here Also you have X and a Y in the training site Access your independent variables wise here dependent variable So like any of the algorithm this also tells you what is the relationship with an X and Y How ex gets transformed into a way right to that extent.
It’s similar to the other algorithms but this algorithm is based on the observations on how human brains work on A lot of research was going on in this field on biology Neurology to be precise we’re back in the 19 for peace Even today is going on But today any results that is going on in definitely little It’s not related to the research going on in biology but in those days these were tightly coupled A lot of information lot of ideas with border from the world of biology into this However today’s research whatever we’re coming across today they have nothing to do with politics So you don’t need to worry about not knowing biology if you don’t know biology The Iwas in those years 19 forties.
How does the brain work Was a big question even delay Today we don’t know how it works but it works It’s one of the most powerful computing systems we have If you read all these research papers that saw him it’s so surprising so mystifying How does brain really book We don’t know Okay during those two years when the research was going on there trying to understand how this not a little brain the neural network in the brain works the natural neural network How do the neurons were hard Is it hold this process The information on During that time some biologists who were also mathematicians they tried to replicate the findings over there and see if we can make the existing computer systems off that time smart are intelligent enough to mimic the functioning of the brain human brain.
We can make this comfort assistance Learn to do things without being told how to do When an infant stands up and walks for the first time it learns how to walk for the first time on its own right so that Iwas can make a computer system learn to do things without the necessity off training it how to do it without telling it how to do it Now To appreciate this you have to go back to the 19 forties where there’s no by phone There was no pass school There was no see There was no job The only language waas machine learning zeros and months.
I’m not sure how many of revoked on machine learning language machine language Sorry machine not machine money machine language which is the language of zeros and ones Even to act two numbers you have to manipulate the registers shift the registers left and right and do all the things this to our two numbers Now imagine asking that computer to do for you We’re very complex task What kind Of course you’ll have to write in that machine Language zeros and ones I know I don’t know how many If you’re seeing this I’ve seen this myself So I’m not used it in 19 eighties when I was a student as just getting into the Khan signs and all Mr have punch cards punch cards with holes in it It’s a punch card would look like an invitation card for managing mutation cards in that you will have Collins.