Artificial Intelligence Tutorial | Artificial Intelligence Course | Intellipaat

Hey, guys Welcome to Intellipaat in today’s world Artificial intelligence has evolutionize how we approach and solve problems It’s not only a skill that you need to have to get a highly paid job, but also to build something really cool Artificial intelligence is an extensive, large field and it’s very difficult to wrap your head around all the concepts that are involved in It is difficult to get pastorally all gone And terminologoes associated with it Keeping that in mind, we have come with his YouTube video on artificial intelligence This video will teach you everything that you need to know in artificial intelligence As a beginner, before you start, please do subscribe to Intellipaat’s YouTube channel Do not miss an update Now, let me walk you to the agenda for today’s video First of all, we start to know what artificial intelligence is, is brief history and introduction Then we move on to key major points in history of artificial intelligence Furthermore, we understand what an artificial intelligence algorithm is, but a super wise or unsupervised And then we go on to the world of deep learning Furthermore, we’d also like you to understand what is an artificial neural network The topology of an artificial neural network and positron and many more then we take a closer look into one of the most popular tools using your AI platform Your tensorflow and your Pytorch Going further ahead, we go into the details of your tensorflow learning concepts like tensor computational graphics feed for neural networks, multi-layer perceptrons and back propagation Last but not least, we make our own facial recognition program using computer vision and almost all the popular tools in ai Now, if you guys want to become a certified artificial intelligence professional, we recommend that you gleed through the description that’s below We at Intellipaat provide all courses of artificial intelligence covering up all major concepts Thank you very much Now let’s continue the session One important aspect comes into picture when we talk about artificial intelligence It is the year 1956 guys Well, pretty much In the year 1956, the term artificial intelligence was actually coined Well, how did this actually come to happen? I mean, not a lot of us might actually know that it was coined at the Dartmouth conference and so much more But what exactly is this conference? Well, in 1956, for a couple of years, basically, they used to happen You know, that used to be meetings and a couple of meetups where a lot of scientists and researchers used to come together under one roof And this eventually was called the Dartmouth conference And you’re basically what started out as a summer research project And it lasted about eight weeks At the end of it, a the product of their research and the product of basically their entire project was one person who coined out the name, what is artificial, you know, artificial intelligence And it’s ever been stuck with us right back from 1956 Right I hope you got the Lot of the Rings reference there So it’s a one conference to rule them all It’s basically wondering to rule them all if you haven’t got that And coming back to the presentation So even after 64 years of the term being coined by Mr. John McCarthy, he is regarded as the father of artificial intelligence guys Even after 64 years, the world is still going mad, still going gaga about artificial intelligence when you think about it Well, especially in the last 10 years, right I mean, everyone is on the you know, everyone is on the trend of artificial intelligence And it’s just gonna go like this for the next couple of decades is actually what analytics allows guys So here’s a quick go excerpt that I read So pretty much is from the person called as Daphne Koller She She’s a scientist at the Stanford Artificial Intelligence Lab And she says that, you know, John McCarty believed in artificial intelligence in terms of building an artifact that could actually replicate human level intelligence in replicating a human being Human is pretty much the toughest thing that even even these big scientists could ever think of And this was thought about We’re back in the 1950s as well And she says that he was very happy with a lot of artificial intelligence today because, you know, at the end of it, it does not replicating the human level intelligence that it was actually sought out to be And it ended said that he wanted artificial intelligence to pass something called the Turing Test Well if you guys know about the Turing Test, just head to the common section and you know, talk to your fellow viewers and let them know about the during test And if you cannot hold on to the curiosity, give me two or three more minutes I’m going to tell you what the Turing Test actually is, guys So on that note, we need to check out what artificial intelligence actually is So let me give you the break up of the textbook definition, guys Well, the definition goes like this It’s the theory and the development of computer systems, basically No Which are able to perform tasks normally, which actually require a human to do the task It might be something as visual perception, speech recognition, decision making and translation between multiple languages as well Well, that definition will make sense if you actually go through twice I mean, this is the textbook, right? So I usually keep all the sessions very simple for all of viewers, from the beginners to

the advanced level guys So, you know, you can have a lot of take away from it if we actually, you know, go about simpler, simplifying certain things for you So the current goal of artificial intelligence, if I have to put it in one picture, is basically to tell you that it does does the mathematics on steroids guys, because when you think about it, all artificial intelligence is trying to achieve is better ways and faster ways to do something that a human can do a lot of for like machine learning, deep learning and come into the picture when we talk about mathematics as well, because at the end of the day, to achieve artificial intelligence we use, so many concepts, which involves a lot of mathematics and implementing this mathematics using codes guys And at the end of the day, it’s mathematics plus programming, which will actually lead to us achieving cognition guys So what about the future? Well, again, achieving cognition has always been, you know, the goal even of the you know, it’s been the goal of the person who actually coined the term as well And pretty much every one of his followers and everyone wants to do something with artificial intelligence, pretty much can tell you about how important and how amazing it would be to actually achieve cognition Well, by cognition, I actually mean that getting a machine to replicate what a human can do exactly the way the human does or even do it better as well Guys So what do you guys think about it? What do you think in the future? I mean, what do you think? Artificial intelligence has a grip on the future So head to the comment section and do let me know that So then you might have heard of these terms, as I mentioned Right Machine learning, deep learning and so much more Well, guess I would love to explain all of these to you guys But we’ve already put out a lot of videos on our YouTube channel and we have amazing subject matter experts writing our blogs for us as well So if you want to know more about these Terms machine learning, deep learning or how they are used to achieve artificial intelligence in general, I suggest you head to our YouTube channel after this video And then you can go about checking out blogs as well But hey, you guys, make sure you stick to the end of the video because I’ll help you fast, like your career to become an artificial intelligence engineer And I’ll be and I’ll be giving you a special coupon just for the viewers of this video guys So on that note, let us quickly check out the timeline of what artificial intelligence was, what it does nowadays Now, if you guys want to become a certified artificial intelligence professional, we recommend that you lead to the description that’s below We at intellipaat, provide all causes of artificial intelligence, covering up all major concepts Thank you very much Now let’s continue with the session Let us get Back in nineteen forty three Well, well, it was a time of World War Two guys It was almost ending And pretty much this person is the reason in my opinion, why World War Two ended I am a major history buff And I believe personally and many people agree with me Thousands of people agree with me when I see it as Mr. Alan Turing, who was a British mathematician, was the person who ended the World Wars because the Germans used a very nice coding system, which was quoted as the enigma And, you know, they used to use that to send messages And while the allied forces never got to know what these messages were because they were all encoded and they just look like rubbish But Mr. Alan Turing developed the first artificial intelligence machine He went about using a lot of mathematical formulas You know, it took it took a lot of days, months and years to go about cracking this amazing and encoding system, which was basically generated by the Germans to go about sending their messages, guys There’s also a movie on that So if you could all watch that, it would be pretty nice as well So this person alongside breaking the enigma also it is called the Turing machine, basically And then he invented something called as a Turing test Well, what does that during this do? Well, I just told you three slides back that would explain it Right So here it comes So basically, consider the scenario that you are a human and you you obviously are humans, right? So basically, you’re sitting in front of your computer and you’re asking a couple of questions to random people around you an instead of saying it as random Let us consider two situations One, your questions are being seen by a computer and the other, your questions are seen by another human So whenever you ask these questions, if you cannot figure out the difference between the answers that the computer can give and the human can give I mean, if the answers are equally good, well, you can not tell apart between the computer answer and the human answer, though, that particular computer with answers for you has passed the Turing Test guys so that computers intelligent enough to know pretty much give you answers just like a human, and it can do a damn good job at it, guys So this particular test, this sets the bar for, you know, an intelligent machine, let’s say, because if a computer can fool someone into thinking that they’re actually talking to another person, then we can say, you know, to a certain extent that pretty much that computer has achieved enough cognition because it is pretty much going about, you know, let’s not say pretend, but it’s trying really hard to replicate a human and it’s doing a good job at it as well guys so fastrack to 1956 Again, the term was coined that And scientists are scientists, you know, begin to tackle the concept

They’re like, OK Another Term said, let’s work on it And one of the top scientists, his name is Mr. Marvin Minsky So basically his view, along with McCarthys view, got him a very good funding from the then American government, United States government, who basically hoped that, you know, with the help of artificial intelligence, it will actually give them an upper hand in the Cold War, which is ongoing back then guys So pretty much Mr. Martin Minsky, along with McCarthy, actually got good funding from the government and they went about doing that as well, guys So this one other important thing happened in 1956 was they basically Mr. Marvin went about starting the artificial intelligence laboratory at MIT guys and this lab right from 1956 all the way to 2019 and 10 years from now, has been punching out amazing solutions to the world’s problems And it has been the most efficient laboratory in terms of artificial intelligence that the world has ever seen And there are very intelligent people and very much, you know, a force in these people which drive them towards achieving artificial intelligence is what they can tell u guys And then again, a quick, fast track to 1981, when 1981 artificial intelligence actually got commercialized So this person, called Ken Olson, he is the founder of Digital Equipment Corporation, guys So basically here he realized that, you know what? As a business leader, we can actually start making use of artificial intelligence and start selling Artificial Intelligence to the common folk as well And then our first track to 1997 This was one of the most important key moment, in my opinion, back in the 90’s regarding artificial intelligence, because this was the first time there was a chess tournament conducted between the Grandmaster, Garry Kasparov He was a world champion multiple times in chess He actually played against IBM’s deep blue Deep blue, basically there Then supercomputer and then it deep blue actually defeated worlds Best chess player at that point of time, guys And people were saying that, you know, IBM’s deep blue could actually think like God because it was evaluating over 200 million moves in the game of chess at once or for every move It could bring up to 100 million combinations of how you could go about playing How amazing is this guys? Is it still 1997? And then, you know, artificial intelligence was already on the roll and people could say it could think like God I mean, that is a huge statement And every time I go or reading about Kasparov or Deep Blue, I am amazed by the fact that this actually happened In fact, multiple tournaments happen You can probably read about it after the video games and then fast tracked to the 21st century We are finally at 2002, guys, I’m sure most of you might have heard of Roomba Well, Roomba is a company pretty much you know, it’s it’s from iRobot It’s basically a carpet cleaner and it’s a vacuum cleaner So it’s a robot vacuum cleaner So, you know, in my opinion, again, this was another big achievement because all it took was just a few layers of behavior generating systems which basically guided this machine around your house, around your living space And it could efficiently clean carpets and floors autonomously without your help And this has been sold across the world as well It has There is 50 million units all and I use I use pretty much roomba every day So it basically has a docking station where it comes out of a docking station Our station, it cleans our living room It goes into the bedrooms, it cleans the bedrooms And pretty much it just comes back Noza at its docking station is and this goes back charges you know, comes back again two times a day You can program all of these And it does an amazing thing It doesn’t even you know, sometimes it starts doing it at the middle of the night And I’m taken aback Basically, you can set all of that events and the timings of which your robot actually starts cleaning and all of that But it’s actually amazing to know that as an autonomous system that I personally use to get my housekeeping stuff done, guys So they have sold about 15 million units of these across the world and the number is rising as well And then coming to then, maybe not so nice part of it in a couple of people’s opinion But in my opinion, again, another amazing use of this isn’t war machines, guys So many autonomous world robots were pretty much are deployed to Iraq and Afghanistan In fact, two thousand plus robots which are capable of war, everything from helping our soldiers carry heavy loads, diffusing bombs, which would be a threat to the life of the soldiers patrolling enemy area You know, all the way till they could make a stable war robot, which is capable of inflicting heavy damage on the on the opposition forces, guys Well, this is good in terms where, you know, our forces or any country’s armies, navy or military forces, are not put into immediate danger Well, if the robot pretty much, you know, is shot dead or whatever, if it’s a malfunction or something or if it’s taken a lot of damage during the war There is no human life lost there And that is the important thing That is a good thing Well, the bad part of it is if this pretty much keeps on going, then every superpower

and every nation with the superpower will want to go about holding these And then, you know, I pray to God there isn’t another big war coming But if there is, then you’ve even, you know, see a lot of these saw or guys That’s the sad part of it And I am sort of confident that really won’t come to that But if it does, then you can see a lot of robots there Guys, let’s pray it doesn’t come to that Well, the next important thing in between the robots and the tesla, the pilot actually came about Google’s self-driving cars So I’m not sure if you guys know about Google’s self-driving car But instead of just telling you what the self-driving car is, I just thought of telling you the Tesla to pilot directly, guys You’re basically this was the biggest breakthrough in terms of artificial intelligence technologies, because Tesla, as a company in the United States, co-founded by Mr. Elon Musk, all started providing cars to the public where, you know, the car could drive itself throughout, it could park itself And all you had to do is basically sit in the garden every fifteen minutes or so You had to hold the standing to make sure that you’re not sleeping or something So how amazing this is, is there is There’s a photo on your screen right now on the top Right If we can’t take it out, that isn’t an actual photo from the car’s dashboard guys So it basically instead of having a huge speedometer or something It has this amazing display that it shows you It detects all the cars around it It detects rain and it detects the shortest distance Guys you have to read about Tesla’s autopilot Sure, many of you guys will find it fancy guys, in my personal opinion, if this is brought throughout the world The capabilities of self-driving, that would be an amazing thing As we’ve already seen, a lot of companies actually bring out parking gadgets and autonomous parking and it’s basically car less parking help or whatever it is A lot of companies like BMW skoda and you know, Mercedes and all these big guys have already had autonomous parking capabilities But when you think about, you know why, when you think about how it goes about in every country, I’m sure that’s very different But then when you when you contemplate about it If this was brought to every country, it would be amazing Right And then coming to 2019, guys, I mean, it’s one more month to the end of 2019 And one amazing thing happened this year as pretty much in the field of medicine with respect to artificial intelligence guys Well, because artificial intelligence, as we all know it has been helping the world in many different ways And it has done so in the in the in the field of medicine as well Well, it can be used to ameliorate, you know, reading time for digital breast promosynthesis disease while maintaining or improving the accuracy of the reading Well, what it’s basically trying to tell us is that, you know, it artificial intelligence is basically helping us to accurately make sure we detect cancer cells early on days So this again, we already know all the victims of cancer and we already know that it is an uncurable disease and it takes a lot of your time, endurance, your and so much more to go go through with this guys But then if artificial intelligence can help us to take out, you know, what cells are harmful, what cells are harmless, basically what is a benign and so much more than definitely this would help, right? Well, it does actually The research has already been done It is being implemented in many countries And, you know, as a part of an artificial intelligence researcher, I am extremely proud of the fact that we’ve brought artificial intelligence down to the field of medicine guys So on that note, we need to check out what’s next We’ve seen all these different fields and I’m sure there are, you know, hundreds of fields that I could have put in here due to the shortage of time and pretty much I know to keep it to the scope of this tutorial for all of our viewers I had to cut it down, guys, but take a second head to the comments section and let me know what you think about the use of artificial intelligence and what are the other applications that you find amazing, guys Now, if you guys want to become a certified artificial intelligence professional, we recommend that you gleed to the description does below We at Intellipaat provide all courses of artificial intelligence covering up all major concepts Thank you very much Now let’s continue the discussion So on that note, we need to quickly take out one of the types of artificial intelligence guyss If we keep diving into the depths of these, then it might actually go to the scope of this video So let me quickly tell you, what are the types of artificial intelligence when we have three different types of artificial intelligence guys? One is narrow ai Next is generally AI and then super A.I. guys with narrow A.I It’s basically in order the concept which is meant to assist or take over specific tasks and do it better than a human would do Basically an example of narrow ai, I would be pretty much you are your mobile phone assistance I can be Siri, Alexa, Cortana and so much more So you just make your work easier and instead of opening the calendar app, setting your calendar All you do is say, hey, Siri, you know, set my calendar for so-and-so So that again is a very good example of narrow ai guys, coming to General AI It’s basically taking knowledge from one domain and transvering this exact knowledge into another domain Well, Generalb AI will

Or, you know, almost require the intelligence of a human if it does not require more intelligence But you have to match the intelligence of the machine with a human or something Very I’m sorry that Already close to that Well, a good example to give you about this is when you think about it, it’s ah, you know, it’s usually a very advanced level of chatbots I mean, siri alexa All these are amazing chat bots And then generally,AI as again, a lot of mathematics involved here as well A very good example would be AlphaGo and so much more guys So AlphaGo is basically, again, another computer which shall beat the world champion at this game called Go I wouldn’t say it has the cognition level of a human, but pretty much even though it sounds simple on the on the outside, it is actually very complex And even the world champion said that this is the most intelligent go player that I’ve ever met in my life So that brings us to super A.I., guys Super A.I. is on another level This is up, I would say, artificial intelligence, mathematics, programming, all on steroids at once Because when you think about it, a super AI consists of machines that are extremely smart compared to humans They can outrun a human, outthink a human, outgun a human, and so much more for either good or bad reason Super AI, you know, super ai is pretty much either in good terms or bad terms guys I’ve never seen neutral centiments About super AI But for the scope of this video, all you need to know is that Super AI pretty is pretty pretty much the domain which consists of machines that are extremely smart When you compare it to humans guys So on that note, you need to quickly talk about what? What does it cost for all of these in intelligence? So what makes artificial intelligence intelligent? Well, in my personal opinion, the art of training a machine is amazing guys when you think about it You can go about showing your computer images of cats and dogs and asking it to classify it for you I mean, from our childhood you’ve been practice we’ve been trained by either school or by family members or friends to know what a cat looks like and what our dog looks like But your computer does not It might have taken you three years to figure out the differences or one year or two Let’s say in one year, you took one year to figure out the differences Your computer will probably take five seconds or 10 seconds to figure out this entire thing What took you one year, guys? So when you think about it, everything from a very simple cat dog classifier all the way to achieving cognition, in my personal opinion, is just art guys, because at the end of the day, taking a concept which does not exist and bringing it right up to the level of a human being and in in a in a lot of people’s personal opinion, human beings are the most intelligent species out there right now Well, unless there are aliens and you know, all the conspiracy theories Well, jokes about coming back So, you know, bringing bringing a machine up to the level of a human is definitely achieving cognition And it’s being done right, guys So we need to think about the challenges we face when we go about making artificial intelligence Intelligent guys are therefore important points which are posed as challenges One is knowledge guys because again, when you think about it, when you have to train a machine to figure out stuff on its own and hold these and implemented later is very difficult This brings us to the second point while holding it If we think holding getting a machine to hold information and to make sense of it is difficult, then think about learning because you can learn at a rapid pace, at a slow pace or whatever it is But getting your machine to learn Getting a machine to understand about a new concept Contemplate about it To work with it, analyze it, store it and use it later as something very difficult And this brings us to the next thing, which is problem solving Even though you got your machine, to have all the knowledge and learn very easily needs to go about implementing all of these up You know, in an imperfect way, guys, when it can know, it can have the knowledge of all the world lead things, which goes by But what is the use if it cannot solve a problem? Because again, all the artificial intelligence terms are whatever points towards artificial intelligence mostly is a problem Right Anything in the field of military medicine or so much more We give the developers, the researchers, the scientists a problem so that they can solve it for us easily Using artificial intelligence right your machine cannot solve a problem than it is pretty much no good guys And this is actually very tough to achieve as well And next comes obviously cognition, as I’ve been telling you for so many times right now, because again, bringing a machine up to the human level is again a very daunting task Guys, and in the middle of this, let me give you a very quick, ah, fun fact I’m sure most of you guys might have heard of it already Artificial intelligence has been used in terms of litigation You litigation and law and court and so much more as well, how? Well, I’m sure you guys have heard of JP Morgan Right They are the banking giants of the world

And they get they came up with an artificial intelligence model, which basically, you know, was used to go through and analyze some legal documents You wouldn’t believe the numbers in the first time I read this A couple of months ago, I was taken back, because when you think about it, it’ll reduce the workload from 360000 man hours to a matter of seconds Think about it Three hundred and sixty thousand hours of job What? This is what it takes a human to do it But as soon as they implemented that model, it just took the model a couple of seconds to do it And I have read that the model is very accurate as well So when you think about it, think about the odds, 360000 hours and a couple of seconds So, you know, if you do know this fact, then I hope you find this very entertaining And I hope you guys feel very interested about this fact at the same time, because we really know artificial intelligence is being used in every field, but not many of us know how it was actually used in litigation I just found it very interesting to tell us about that And this brings us to the dark side of artificial intelligence Well, is there a dark side? Well, you remember Elon Musk, the founder of tesla, the founder of Space X, the founder of SolarCity, and so much more He’s basically a billionaire entrepreneur When he says that know at least when there’s an evil dictator, that human is going to die But for an artificial intelligence being, there will be No death It will live forever And you will have an immortal dictator from which you know you can never escape So I am sure our views on this particular topic are, you know, we can debate about it all day, all night, and we can have very different opinions about this And we need to respect all people with very different opinions about us, because at the end of the day, all of these are what we look at as different opinions actually come under one umbrella and it helps develop artificial intelligence in the right way, guys So I hope there is a good you know, I am in fact, I am sure that there is a good way and a good path for the development of artificial intelligence But you can you can never know at any point of time Well, according to a lot of experts, that it might take a turn for the bad as well Again, let’s hope that it doesn’t go there And on that note, we quickly bust a couple of myths and facts case So the myth is that superintelligence by the year 2100 will be inevitable Well, when you think about it, it might happen right now It might happen in a decade, in a couple of years So no audit might never happen at all To achieve superintelligence again, as we’ve been talking, is a very, very, very challenging task And the myth is that, you know, only are people who do not know about Luddites are basically people are not very well informed that only these people are worried about artificial intelligence and all of its bad things for the future? Well, not exactly, because even many top artificial intelligence researchers are concerned about how it’s going to go about in a good way or in a bad way as well guys another methods that artificial intelligence is turning evil or just turning conscious and so much more well the actual worry that we need to go about doing is, you know, artificial intelligence actually turning to be very competent with the goal that would be misaligned with our particular goal So it might take a very staunch deviation from what we said it ought to be and it could do something on its own So that is the actual worry when you come to the myth and then to bust another couple of myths, people say that robots are being the main concern, you know, like those terminator movies and whatnot Well, again, here as well, Miss Misaligned intelligence is again the main concern because we live in a world of technology Every device, every home, I’m sure most of the world is connected through an Internet connection Right So you will not need an actual robot incase if something goes wrong as well So that is something to think about And then artificial intelligence cannot control humans is a myth Well, intelligence actually enables control So, you know, we do this very sad thing where we trap animals and gages and create zoos around them I mean, I’m not much of a zoo fan I. I hope I do not offend anyone regarding that So, you know, we are smarter than tigers and all of these other animals so we can go about controlling them because we are smarter Now think about a machine which is smarter than humans It can go about controlling us as well Right So that is a fact to think about again So machine cannot have goals Well, well, well, well This is wrong because again, let us say two planes are fighting in the air It’s called the dog fight or two planes are fighting And one, that, the plane behind just launched a missile If you have seen in a lot of movies that I’m sure you would have, this missile exactly knows where the plane is and to go hit it This basically works on the concept of heat-seeking Well, it has detectors in the front of the missile, which would find the hottest thing it can see And it does go hit it So the hottest thing in a plane is on the plane, which is in front of its engines So this will basically change the plane’s engines and hit the engines

So that is a heat checking missile for you And then the next mythical widdy is at superintelligence is just years away Well, not really It is a couple of decades away, as I pretty much told you But again, it might take way longer than two decades, three decades, four decades and so much more Do you know, just to make it safe for us to use this? Well, guys So that is all the dark side concept that they had to just inform you guys in case you weren’t knowing So this brings us to the next concept, which is basically the domains in which artificial intelligence has its hold Again, no. one, as I’ve been saying for a long time, is healthcare Right? That is the best thing that I’ve seen, artificial intelligence being put under so many things, live as well So it has been so certainly integrated into your lives that you would not even know it is there when you think about it now, you know, probably this open a health app, health app installed on your mobile phones or your computers, you’ll be talking to a chat board and sort of a human And youwouldn’t know it most of the time You can just set an appointment with your doctor and ask more information about your appointments and so much more by just talking to a machine instead of talking to a human on the other side, guys And from that, helping detection and curing of cancer is again an amazing thing, in my opinion And the second thing is stock trading guys as an investor Personally, I could sort of relate on this as well, because if I stock trading again isn’t A very topsy turvy world, right, so sometimes you lose money, sometimes you make a lot of money But if you had a way if you had a different approach, if you had a more intelligent being to look at graphs, to look at data, to make better sense of it than an expert trader, then that would be an amazing thing, because you could end up making more money and losing less money at the same time Right So this brings us to one other next thing, is sales guys Well, this is this is a personal recommendation from me because here Intellipaat let me tell you how we implement machine learning and artificial intelligence for our sales team So we get about 10000 or 15000 leads, sales leads every single day Imagine calling those 10, 15000 people, guys So it would be a daunting task to go about doing it But we have a machine learning model which is in place, which basically helps filter out these people, you know, categorize them and make them make sure it basically makes sure that they get in touch with our course advisors and then later our course advisors It’s basically help them get the course that they would actually looking for guys So this makes the entire process of going go through the part of sales a very interesting, a very simple And we get lot of analytics from that as well, guys So this is us at Intellipaat using artificial intelligence to help out you learners as well, guys So that brings us to my final thoughts regarding artificial intelligence So you might have this question at this point of time, will artificial intelligence take over the world? Well, when you think about it, since you are the most intelligent, I mean, since we are the most intelligent being right now, it is as good as we are, because if we are extremely good at programming, our artificial intelligence model will be extremely good at programming So it totally depends on us and how good we are on less machines Start writing code and they can ated better than us So until we go that we are the most intelligent beings we know and that artificial intelligence will be as good as we are guys So let’s actually get a bit deeper and understand what is intelligence? So intelligence can be defined as one’s capacity for understanding, one’s capacity for self-awareness, one’s capacity for learning and one’s capacity for problem solving That is how well is something or someone able to understand how well is someone able to learn new things and how well someone will do solve problems by themselves So now that we know what is intelligence, let’s understand what is artificial intelligence So when you apply the same intelligence to machines, this is known as artificial intelligence Let us imagine there’s a machine which can understand things which are normally understood by humans There is a machine which is self-aware and there is a machine which can solve problems by itself That’s just amazing, isn’t it? All right So this is the artificial intelligence, which I’m talking about So, no, it’s not Let we also know what is intelligence I’ll ask another question So tell me, what is it that makes humans intelligent? Well, we as humans can reason We as humans can learn We can perceive We can solve problems And we also have linguistic intelligence That is, we can figure out what is someone else seeing And we can also understand the grammatical intricacies of different languages

So, again, my question would be, what if a machine could exhibit all of these factors normally shown by a human? Again, that’s just amazing, isn’t it? So this is what is known as artificial intelligence So a machine which can sure read, normally shown by a human that is known as artificial intelligence All right So now that clear artificial intelligence, let’s segregate AI, ML, DL So normally most people get confused between artificial intelligence, machine learning and deep learning So this is where I’m going to help you out in understanding the difference between these three So we have AI at the top and you can consider machine learning and deep learning to be subsets of AI So again, machine learning and deep learning are just ways to achieve artificial intelligence So I’ll restate it again Machine learning and deep learning are just ways to achieve artificial intelligence Now, machine learning is that part of artificial intelligence, which aims to teach the computers the ability to do tasks with data without any explicit programming Right So we don’t need to do Any explicit programming and algorithms do tasks by themselves and in ML we mostly use numerical and statistical approaches to achieve artificial intelligence And then we have deep learning, which is actually a subset of machine learning So first we have A.I. and then we have ML and then we have been so deep Learning comes in there, machine learning fails and we apply deep learning through something known as artificial neural networks about which will obviously learn later All right So now let’s understand artificial intelligence and a big asset So as I’ve already told you, artificial intelligence is the superset under which comes machine learning, under which comes deep learning and then machine learning and deep learning are basically ways to achieve artificial intelligence Now, these are the different areas of research, of artificial intelligence So you have ML, again, a part of ML is deep learning Then we have natural language processing all here We basically understand what this spoken or written by a human And then we have speech where we either translate this speech to text oe we would translate text to speech The next subfield is robotics and then we have autonomous vehicles under robotics So Google self-driving car is an example of this over here Now that we’ve also understood the difference between artificial intelligence, machine learning and deep learning, let’s see different examples of machine learning around us So most of you would have shopped on Amazon Now when you go into Amazon, you see that there are some products recommended to you know how do you think that would happen? So this is something known as recommendation engine and recommendation engine is nothing but a component of machine learning So let’s say you and your friend buy similar products to your friend buys five products and you buy three products Now are of those where there were three products you buy are the same as what your friend buys So let’s see, the common products are an iPhone, a back go for the iPhone and a Bluetooth headset Now, let’s say the other two products bought by a friend would be a MacBook and a mouse Now, since there are three products which are seen between you two This is why the products which our friend has also bought, those are the products which will be recommended to you So on the basis of the commonality between you and your friend, you will be recommended a Mac book and a mouse as well So this is nothing but a concept of machine learning And then we have Amazon Alexa So Amazon Alexa as Amazon alexa is a really good example of speech recognition You know, when you say, Alexa, turn on the lights, it look down on the lights When you say, Alexa, look right for me and I’ll do exactly that When you say Alexa order a cheez pizza and that is exactly what Amazon Alexa will do Now, Alexa is just a machine right but when you see it do something Ordered a pizza book a cab for me Turn on the lights You know, how is the machine able to understand all of this? So the idea behind this speech recognition and that is, again, a component of machine

learning And then we have Netflix movie recommendation So let’s say you watch two TV series, first TV series as friends And the next TV series is Big Bang Theory And since you watch these two TV series which belong to the gener comedy, that is why maybe you will be recommended How I Met Your Mother or you can be recommended Silicon Valley or some other Tv series belonging to the comedy gener So this again, is machine learning And then we also have Google traffic prediction Let’s just say see are traveling in your car and there is huge traffic There you are And you desperately want to get the traffic So you done on Google Maps and Google Maps tells you the best direction From there, the traffic would be the released know how it is Google Maps do this This again is machine learning now that we’ve looked at different real world applications of machine learning Let’s actually understand what exactly is machine learning So as I’ve already told you Machine learning is a subset of artificial intelligence which gives the machine ability to learn without being explicitly programmed over here Data is the key Or in other words, you basically teach machine how to learn without any explicit programming and the machine learns with the help of data So now that we know what exactly is machine learning, let’s also understand how this machine learning work So as I’ve already told you Machine learning depends totally on data So first have taken the data set & Divided into two parts The first part would be the training set And the second part would be the testing set and they will train the model on top of the training set So once we train the model, we will give it a new data and check for its accuracy On top of that new data and the accuracy of that new data comes out to be good enough Then we will go ahead and use that machine learning model On the other hand, the model which you build, the accuracy of that model is not good enough Then we will go ahead and fine tune that model till we get the desired accuracy But this is the basic premise behind machine learning Now let’s forget the sub categories of machine learning So we have supervised learning, unsupervised learning and reinforcement learning So when supervised learning You can consider that the learning is guided by each other So we have a data set which actually acts as a teacher and its role is to train the model or the machine So once the model gets trained It can start making a prediction or decision when new data is given to it So let’s take this example So what? Here we are training this machine by giving good samples of data So what? Here the data is nothing but different images of Apple and along with each image of Apple We are also giving it the label of the image Right So this image goes with its label, which is Apple Again, this image goes with its label, which is Apple Again, the same with these two Right So we are teaching this machine that whenever it sees an image, something like this, it is nothing but an apple And after time, when we give it a new data from what they were learning, it has done It will predict whether it’s an apple or not So on the basis of its learning, this machine predicts that there is a good possibility that is actually 97 percent possibility that the image, which has been fed to the machine, is nothing but an apple So we’ll use case of supervised learning could be spam classifier So spam classifier basically means that whether the email which we get, it’s a spam or not And that is done on the basis of different textual parameters So let’s see A genuine email would contain too many exclamation marks It didn’t contain a catchy headline and so on But on the other hand it to a spam email Maybe it would contain a lot of exclamation marks with maybe a load of numbers and ill have statements like Hey, congrats, you won a lottery or hey, could you help me out for this so this spam Classification is basically an example of supervised learning

Then we have unsupervised learning So an unsupervised learning the model learns through observation and find structures in the data So once the model is given a data set, it automatically finds by turns and relationships in the data set by creating clusters in it? So what it can do is add Lables to the cluster Like it cannot see this is a group of apples or mangoes, but it will separate all the apples from mangos So over here we have this set of images Now, this unsupervised learning model which is applied on this, it is segregate these fruit on the basis of similar characteristics So over here we have segregated these four into one cluster, these three into second cluster and these three into third cluster Now, even though the unsupervised learning does not have any labels, it has still segregated these three in three clusters Right So the machine or here does not know that these are apples These are oranges or these are bananas Yet it has segregated these three on the basis of similarity of characteristics So it found out that these four objects are similar to each other And there is quite a bit of variability when it comes to these four objects and these three objects Similarly, this machine was able to figure out that these three objects are quite similar to each other But when compared with these three objects, they are very dissimilar This is the underlying concept of unsupervised learning and a good example of unsupervised learning would be again Netflix movie to recommendation So you would hear the movies are segregated on the basis of different geners So over here Tv series These like friends how I might do mother and Silicon Valley are clustered in one group because those come into the same category Similarly, movies such as Secret Superstar and Dangal could come under the same degree because they have the same lead actors So Here we are segregating the movies on the basis of similar characteristics, even though there are no labels in it And it’s finally time for 3rd machine learning type, which is reinforcement learning So or here there is an agent and there is an environment and the agent interacts with the environment and finds out what is the best outcome for it So it basically follows the concept of hit and trail method the agent is rewarded or penalized with a point for a correct or a Wrong answer And on the basis of positivity, award points gain the model trains itself So let’s take this example So over here, this self-driving car would be our agent and the road, is the environement And this car is interacting with this environement So it will absorb the environement and it has two choices or here So either to go straight or turn right Now, let’s see this agent or the self-driving car decides to go straight Then what happens is it goes straight into this barricade So then it realizes that the action taken by it was not in its best interest, and that is why it is penalized So since it is penalized, it realizes that the action taken by it was wrong And that is why from the next time onwards, it will do the opposite action So instead of going straight, it’ll take right turn And when it takes that right turn, it realizes that that road is correct And the agent is rewarded So this is how reinforcement learning basically works for the agent, interacts with the environement It takes an action And if the action turns out to be incorrect, it does penalize end of the action turns out to be correct It is rewarded So this cycle goes on and on till it completely learns it’s environement properly And best use case of reinforcement learning is again self-driving driving So companies that does tesla and Google are working on this self-driving cars So just to sum it off, these are three different types of machine learning algorithms So we have supervised unsupervised and reinforcement machine learning So under supervised we have regression and classification And in unsupervised we have clustering techniques, association analysis And hidden marcoh model And then the third is obviously reinforcement learning, which works on trial and error method And if you want to do some really cool machine learning

Projects, you can check the site out Now let’s look at some limitations of machine learning So when it comes to machine learning algorithms, they would require massive stores of training data So again, as I’ve told you Machine learning is totally based on data which it has So if you have more amount of the data, then it will be able to give correct accuracy So let’s say you take in very small amount of data and there’s a good possibility that the results which you’re getting are very biased of very incorrect and also error diagnosis quite difficult when it comes to machine learning because again, the amount of data is very huge And wherever there’s a mistake, you would have to go through that entire algorithm, which you’ve written, and then find out that particular mistake by yourself, which is very difficult Also, when it comes to machine learning algorithms, they’re not really that creative So these ML algorithms are built only for one specific purpose So let’s say I’ll build a machine learning model which will predict whether it will rain or not today Now, if I want to use this same model to predict the stock prices will not work Right So basically one model is built only for one particular task So this is the lack of creativity that I’m talking about when it comes to machine learning Also, there are a lot of time constraints as the model has to learn a lot of historical data . So that was everything about machine learning Now let’s start off with deep learning So deep learning is a subset of machine learning where it learns through the data representations as opposed to task specific algorithms So we saw that the drawback in machine learning models was that the models are specific to only one particular task But this is not the case with deep learning models as these deep learning models are based on the data representations and these deep learning models are mostly built with something known as deep neural networks So this is how a deep neural network looks like So these deep neural networks completely learn that data which is fed to it So this is the data So let’s say if this image of a woman as fed as the data to the deep learning model, then it will completely extract all the features of this data by itself Again, the difference between ML and deep learning or here is that the extraction, the featured extraction in machine learning is manual But when it comes to deep learning, the feature extraction is automatic So the deep learning model automatically extracts all of the features associated with that image And when new images are fed to it, it automatically is able to tell whether the images seemed to this or not So this or here we have a graph which tells us how this performance vary with respect to the amount of data So what happens in machine learning is that as we keep on increasing the data, the performance increases only up to a particular threshold After that, if we increase anymore data, that is no increase in performance So this is another problem.when It comes to machine learning, but on the other hand, when it comes to deep learning, the more amount of data you give it, the Better will be its performance And that again is because deep learning is based on learning data interpretation So the more data you give it, it will automatically learn all those features of the data by itself And it will be keep on increasing its performance gradually Now let’s look at some applications of deep learning so speech recognition as one application of deep learning Now you need to understand that you cannot build speech recognition applications with machine learning So this is where machine learning fails and deep learning comes in and helps you to build speech recognition applications Also, another application of deep learning self-driving cars So we see or hear that the person is just sitting here, is not even touching the steering wheel and the car is driving by itself So just an amazing application of deep learning And then we have language translation over here So this, again, is a part of deep learning So here we are typing something in Spanish And it is being automatically converted in to English So we also have visual translation over here over here, this text or this board as in some

Random language and this app or here, which uses deep learning, automatically converts this visual in to English So those were some applications of deep learning Now let’s actually understand how it is deep learning work So most deep learning methods use neural network architectures, and that is why deep learning models are often referred to as deep neural networks So a deep neural network basically has these three modules and input layer the hadden layers and the output layer and the term Deep usually refers to the number of hidden layers and the neural network So traditionally neural networks only contains two or three hidden layers while deep networks can have as many as 150 hidden layers Now that’s a very huge amount, isn’t it? So deep learning models are trained by using large sets off label data and neural network architectures that learn features directly from the data without the need for an annual feature extraction So all of the input data is given to this input layer, and this input layer automatically extracts the features by itself Now that data is sent to this hidden layer which performs all sorts of processing tasks and then the final result is given out to the output layer So now let’s also understand what exactly is a neural network So when neural network are computing model, who’s layered structure resembles the network structure of neurons in the brain with layers of connected nodes So it can learn from data which can be trained to recognize patters classified data And forecast future events So the neural network is based on the biological neural network of our brain So that is why it does give given the name neural network So the layers are interconnected where nodes or neurons with each layer using the output of the previously layes as its input So its main function is to receive a set of inputs perform calculations and then use the output to solve the problem Now, as I’ve already said, these artificial neural networks are based on something known as a biological neural network So our biological neural network has dendrite cell body and acts on dendrites are what the input is taken Cell body is where the processing is done and axon is where the message is transferred to other neurons And the same thing happens in artificial neural network as well So first we give in the data and that data is processed and then the final processed result is given out as the output So over here, let’s see if we train the data with images of cat and the labels would be either cat or not cat After that, we give in a new image of a cat and then we basically try to predict whether the model correctly classifies this as cat, or not cat And since the model has learned that data properly, it correctly classifies this image

as cat Now there are many ways of netting nodes of a neural network together And each weight results in a more or less complex behavior Possibly the simplest of all apologies is the feed forward network So when feed forward neural network signals flow in one direction without any loop in the signal paths and typically artificial neural networks have a layered structure The input layer picks up the input signals and passes them to the next layer known as

the hidden layer And there can be more than one hidden leader in a neural network And at last comes the output lever that delivers the result Now the first question to pop into your head

would be what is the inspiration behind these artificial neural networks? Well, the answer to that is the biological neural network of

a brain