What is Deep Learning | Deep Learning Simplified | Deep Learning Tutorial | Edureka

What is Deep Learning | Deep Learning Simplified | Deep Learning Tutorial | Edureka

November 20, 2019 49 By Stanley Isaacs


Hello, everyone. This is Saurabh from Edureka. Welcome everyone to
today’s session on what is deep learning? We’ll understand how
deep learning emerged that is what are the limitations
of the previous technologies that led to the evolution
of deep learning. So let us have a look
at the agenda for today. First will understand
what is artificial intelligence, and what exactly is
machine learning, then we look at various limitations
of machine learning and how deep learning solves
or overcame those limitations after that will understand
what exactly is the blurring and we’ll also look
at various applications of deep learning. So I hope you all are clear
with the agenda. Kindly give me a quick
confirmation by writing down in the chat box or if you
have any questions or doubts, you can ask me. Alright, so we
have no questions. So let us move
forward and understand artificial intelligence. But the first thing
that we need to focus on is why artificial intelligence, why do we need
artificial intelligence now, let us understand this
with an example. So nowadays if you have noticed if your car exceeds
the speed limit, so you’ll get a little
It’ll basically a challenge at your home. How do you think that happens? Do you think
that there is a person who is sitting in a chair and actually noting down
all the number plates that crosses the speed limit. Well that is not possible because there might be
millions of cars that pass through that road and at once they might be many cars are to be passing
through that road. So for a human being to actually do this task
is next to Impossible. Now, let us see another approach
to this particular problem. So what we can do we
can actually make use of cameras that will click
the picture of the car that exceeds the speed limit and then we could convert
that picture into a text. For example, we have U KP late-20s way
the human error the risk of human error has been reduced
and at the same time machines, they never get tired. So because of that you can
capture all the images of cars that actually crosses
the speed limit similarly you can think of a many
other examples as well. It is used in order
to recognize a sign that is in Banks. You want to authenticate whether that person
is the bank customers or not apart from that it is Was
for self-driving cars as well. So in u.s. Around 30,000 people
die every year because of Road accidents so
that can be completely removed. If we use the self-driving cars which is based on the concept of
artificial intelligence and let me tell you guys you
might find it very fascinating that people in MIT are using
artificial intelligence in order to predict the future so you can imagine why
we need artificial intelligence. If you have
any questions any doubts, you can ask me it is even used in places where humans
can’t reach for example, a deep oceans
or navigation and Mars. So in those places, we need machines which are smart
enough to carry our tasks. This is why we need
artificial intelligence. If you have
any questions any doubts, you can ask me or if you have any example to share
you can do that as well. Any questions guys. Alright, so we have no doubt. Let us move forward
and understand what exactly is artificial intelligence now
artificial intelligence. I know the word sounds
pretty complex and there are a lot of Hollywood movies that are based
on artificial intelligence. If you have seen
Terminator or Matrix, all these movies are based
on artificial intelligence, but you don’t need
to worry about it because till now we
haven’t reached that level as they have shown in movies
like The Terminator but yeah, the concept is pretty similar. So basically we want systems
and software’s in such a way that they get imitate
The Human Behavior. Now what happens
in artificial intelligence artificial intelligence is
accomplished by studying how human brain thinks and how human brain learns
decide and work while trying to solve a problem and then we use outcome
of this study as the basis of development of intelligent
software and systems. So our major goal is
to have systems or software that can imitate
the human behavior the way they think the way the decide
the way to solve a problem so in that Look fashion, we want a machines to do that. So this is basically artificial
intelligence in a nutshell. I hope you have
understood the concept if you have any questions, you can write down
in the chat box. Fine guy so we have
no questions here. So let us move forward and look at various applications
of artificial intelligence. So this light basically talks about the application
of artificial intelligence. Now, I’ve listed only three of them but there are
millions of applications. For example, it is used
in speech recognition. So whenever you search
something on Google, so you just tell Google
and searched for you similarly. It is used for understanding
natural language as well as for image recognition as well. And there are many
many other applications in which artificial intelligence
Finds Its use for example, it can be used
in self-driving cars. It can be used in CD
for recommending some products. And even when you go
to websites like YouTube or Pandora YouTube knows
which video you want to NEX Pandora knows which song you want to listen to How
do you think this happens? It happens all because of
artificial intelligence. So all of these
are a few examples of artificial intelligence, but nowadays it is used
almost everywhere guys. Trust me on that. Now, let us move forward and understand how to achieve
artificial intelligence now in order to achieve
artificial Legends, there were few technologies that came first came
up machine learning. Now. There are certain limitations of machine learning
in order to overcome. Those limitations came
a deep learning. Now. Let me tell you guys the concept of artificial
intelligence is not new. It was first coined in 1956, but it was just a theoretical
concept then in 80s and 90s. We were talking
about neural networks. But since we didn’t have
enough computational power so we couldn’t utilize it
properly but in late 90s and 2000’s we started
using the neural networks for machine learning
there in 2006. The term deep learning
was coined for the first time that overcame the limitations
of machine learning and from 2010 deep learning
was used commercially as well. So this was just a small history about artificial
intelligence machine learning and deep learning not to
understand this deep learning. We need to First Look
at machine learning and what were the VDS limitations
of machine learning that led to the evolution
of deep learning. How does that sound guys? All right fine so I can see
a lot of people agreeing to me. All right, cool. So we’ll move forward and understand what exactly is
a machine learning. Now. What is machine learning? So machine learning
is nothing but a type of artificial intelligence or you can say a subset
of artificial intelligence and it provides computers
with the ability to learn without
being explicitly programmed. So you do need to hard-code
your machine for that. Let us understand this
with an example. So we have a problem statement in which whenever you
give certain input we need to determine
the species of the plot. And what is that input
that input will be sepal length sepal
width petal length and petal width. So wherever we get
these four parameters or these four variables are
machine should be able to predict what sort
of a flower it is. Now. How do you think
that will happen first what we need to do. We need to train our machine
on the basis of the data that we have. So in this data, we have sepal length sepal
width better length and petal width and we have
species so our machine. Learn from this data, it will determine what should be the length
and width of the sepal and petal in order
to classify it as setosa or were secular or other species
of flowers as well. Now what happens next so
you have trained your data, so you have trained your machine
from the data set, then what happens whenever
you give a new input to this particular machine, it will predict the specie
of the flower. So these are
machine learning works. It is nothing but machine
learning in nutshell. So basically I’ll just
revise it once more so you have a data set. So you split that data
into training and testing data. So what happens with the help
of training data you train your particular machine and after that you test it in
order to determine the accuracy and once it is done whenever
you give the new input it will predict the outcome
or the desired outcome. So this is how machine
Learning Works guys. If you have
any questions any doubts, you can write it down
in your chat box. So we have no questions here. Let us move forward and understand various types
of machine learning. So the first type is called supervised learning now
in supervised learning what happens you
have input variables X and an output variable Y and you can use an algorithm
to learn mapping function from the input
to the output now, let me simplify it for you. So what happens in supervised
learning the data that you have already contained
the classification now, let me talk about
the previous example itself. So from our data set, we knew that if we have this
with this length of our sepal and petal so that will be
the specie of flour. So the classifications
are already defined so that will be
under supervised learning now. Let me tell you
how it actually works. So you have data
you divide that data and training data as
well as test data. So on the basis of this training data
you train your machine and after that you
create a model so as you can see that this page is called
training phase and after that you create a model now. In order to check this model
to get the accuracy you have test data. So you’ll pass that test data
and you will see the accuracy that is nothing
but the actual output – the output that is present
in the test data. So with that you
can get the accuracy, so this is nothing
but a supervised learning and if you have any questions, you can ask me right
now any questions guys. So this is all
about supervised learning. We have no questions. So we’ll move forward and
understand unsupervised learning now in unsupervised learning
unlike supervised learning. You don’t have
any predefined classes. So what happens you have data? So on the basis of that data, you try to create your own class
you try to make sure that whatever class you create
has high intraclass similarities and have a low
inter-class similarities. That means if I
have created to class like this class 1 and Class 2. So the elements of this particular class
should have high similarity, but at the same time it
should have low similarity with the elements of class to so you can think of examples is well of unsupervised
learning for example, if I have a data
about my customers, so if I have a website
and there are millions of visitors and my website
and I want to make sure that I group people
on babies criteria, for example, I can group people on the basis of willingness
to purchase the product that is there on my website
or where they’re coming from. What is the source
all Those things I want to group my customers
and I want to make sure that I have certain high
priority customers and I have low priority customers and I have medium
priority customers. So with the help
of unsupervised learning I can actually do that. I can make a certain classes
of people on whom I should focus more on as
compared to the other class. So this was just
an example guys, you can use it in various
other fields as well. So in marketing this is how you
can use unsupervised learning. If you have
any questions any doubts, you can write it down
in her chart box. So we have
a question from Ashish. He’s asking can you name a few algorithms of a supervised
and unsupervised learning? Sure Ashish and supervised learning you can use
of KNN algorithm that is k-nearest neighbor. You can use logistic regression. You can use decision tree. There are many other examples
of supervised learning and unsupervised learning you
can use k-means clustering. So I hope this
answers your question. All right, so he’s
a pretty satisfied with the answer
any other questions guys. You can write it down. So we have one more
question from Theon. He’s asking what do
you mean by that? The information is neither classified not
labeled or it fine and tell you so obviously I’ll take the same
example of marketing itself. So if I have a few customers that are visiting
my website the data that I have it doesn’t include that these people are
of high priority and these people are
of low priority, right? It is just raw data
about the person who is visiting my website, but at the same time I
will be making those labels or I’ll be classifying them on the basis of priority you
getting the difference, right? So in supervised learning
those classifications or those classes or labels are already present
but in unsupervised learning, I don’t know anything about it. I am going to create
my own clusters or I’m going to create my own classes and I’m going to group people
on the basis of priority. So I hope with the help of
this example you get the point. All right. He’s also satisfied with it fine fine guys
any other questions? All right. So this brings us to our next
type of machine learning which is called
a reinforcement learning. Now. This is reinforcement learning
guides now all happens in reinforcement learning
the machine learns by interacting with space or an environment. So it learns with their experience
with its past experience and also by new
Choice exploration now, I’ll take the analogy of dogs. So if you have any dog
or a pet at your home, so if you have trained your dog
not to get the newspaper and if it gets it then you reward it
with some chocolate or things that the dog likes, right? So the dog will know
whatever he has done. He’s actually rewarded for that. So we’ll continue doing
that but apart from that if he does something else
if it’s one of the newspaper, he brings something else. So what you do you
might even punish it. So because of that
the dog will come to know that it has to get
newspaper every morning. Now the same example is there
in front of your screen so you have this machine so that’s two choices
either to touch the fire or touch the water now first what it does Goes on
and touch the file. So because of that it
gets a burning sensation. Now, it has only other option
that is to touch the water. So when it touches the water
it get some reward. So because of that
it will understand that it does not have
to touch fire ever again. Now, there’s a diagram that is
there in front of your screen. So what happens is
you have an agent. All right that agent
perform some action and on the basis of that action, it will be exposed
to some sort of an environment. Now if that action is correct, then it will be
rewarded with that. But if it is not then
it will change its toys and it will again
perform some action. So this process
will keep on repeating. So this is how
reinforcement learning works. If you have
any questions any doubts, you can write it down
in your chat box. Any questions guys. All right, so no question. So let us move forward and understand when we
have machine learning. Why do we need deep learning
that is will look at various and limitations
of machine learning. Now. The first limitation
is high dimensionality of the data now the data that is now generated
is huge in size. So we have a very large number
of inputs and outputs so due to that machine
learning algorithms Faith so they cannot deal
with high dimensionality of data or you can say data
with large number of inputs and outputs. Now, there’s another problem
is well in which it is unable to solve The crucial AI problems which can be natural language
processing image recognition and things like that. Now one of the biggest
challenges with machine learning models is feature extraction. Now, let me tell you
what our features since that is takes we consider
features as variables, but when we talk
about artificial intelligence these variables and nothing
but the features now what happens because of
that the complex problems such as object recognition or handwriting recognition
becomes a huge challenge for machine learning
algorithms to solve now. Let me give you an example
of this feature. Suppose if you want to predict that whether they’ll be
a match today or not. So it depends
on our various features. It depends on the weather. The weather is sunny weather. It is windy all those things if we have provided all
those features in our data set, but we have forgot
one particular feature could that is humidity and our machine learning models
are not that efficient that they will
automatically generate that particular feature. So this is one huge problem or you can say limitation with
machine learning now, obviously, we have limitation
and it won’t be fair that I if I don’t give you the solution
to this particular problem, so we’ll move
forward and understand how deep learning solves
these kind of problems. Now as you can see that the first line
on your slide, which says that deep
learning models are capable to focus on the right features by themselves requiring little
guidance from the programmer. So with the help
of little guidance what these deep learning
will models can do they can generate their features on which the outcome will depend
on our the same time. It also solves the
dimensionality problem as well. If you have very large number
of inputs and outputs, you can make use
of a deep learning algorithm. Now what exactly is deep
learning again since Note that it has been evolved
by Machine learning and machine learning
is nothing but a subset of artificial intelligence and the idea behind artificial
intelligence is to imitate the human behavior. The same idea is for the Deep learning is well
is to build learning algorithms that can mimic pray now, let us move forward and understand deep learning
what exactly it is. Now. The Deep learning
is implemented with the help of neural networks, and the idea of the motivation behind your own networks
are nothing but neurons what on Iran’s these are nothing
but your brain cells now here is a diagram of neuron. So we have dendrites here
the which are used to provide input to our neuron as you can see we have
multiple dendrites here. So these many inputs will be
provided to earn your own now. This is called cell body
and inside the cell body. We have a nucleus which performs some function after that that output
will travel through eggs on and it will go
towards the eggs on terminals. And then this neuron
will fire this output towards the next neuron. Now the studies tell us
that the next year on now, or you can see the two neurons are never connected
to each other. There’s a gap between them. So that is called. Let’s announce. So this is how basically
a neuron works like and on the right
hand side of your slide. You can see
an artificial neuron. Now, let me explain you
that so what here similar to neurons
we have multiple inputs. Now, these inputs will be
provided to a processing element like a cell body and over
here the processing element what will happens summation
of your inputs and waits now when it moves on then what will happen this input will
be multiplied with our weights. So in the beginning
what happens these weights are randomly assigned, so what’ll happen if I take the example of x
1 so x 1 x W1 will go towards the processing
element similarly x 2 and W to will go towards the processing
element and similarly. The other inputs is wet
and then summation will happen which will generate a function
of s that is f of s after that comes the concept
of activation function. Now what is activation function
it is nothing but in order to provide a threshold, so if your output is
above the threshold, they’re only this
neuron will fire. Otherwise, it won’t fire so you can use a step function
as an activation function. Or you can even use
a sigmoid function as your activation function. So this is how an artificial
neuron it looks like. So a network will
be multiple neurons which are connected
to each other will form an artificial neural network. And this activation function
can be a sigmoid function or a step function that totally depends
on your requirement. Now, once it exceeds the threshold it
will fire after that what will happen it
will check the output. Now if this output is not equal
to the desired output. So these are the actual outputs
and we know the real output. So we’ll compare both of that and we’ll find the difference
between the actual output and the desired output
on the base of that difference we have again going
to update our weights and this process
will keep on repeating until we get the desired output
as our actual output. Now this process
of updating weight is nothing but your back
propagation method. So this is your let Works in a nutshell if you have
any questions any doubts, you can write it down
in your chat box. Now, let me tell you guys this
is just an introductory session to deep learning
just to explain you how it actually emerged. What are the reasons it? Me to existence
and a little bit introduction about how it actually works
any questions any doubts guys, you can ask me so we
have no questions here. So we’ll move
forward and understand what our deep networks. So basically deep learning
is implemented by the help of deep networks and deep networks are nothing but neural networks
with multiple hidden layers. Now, what are hidden layers? Let me explain you that so you have inputs
that comes here. This will be your input layer after that some process happens
and it will go to the next node, or you can say
to the hidden layer notes. So this is nothing but your hidden layer 1
so every node is interconnected. If you can notice after that, you have one more hidden layer
where some function will happen. And as you can see that again, these nodes are interconnected to each other after this hidden
layer to comes the output layer and this output layer again. We are going to check the output whether it is equal
to the desired output or not. If it is not we are again
going to update the weights. So this is how a deep
that work looks like now there can be
multiple hidden layers. They can be hundreds
of hidden layers is back. But when we talk
about machine learning that was not the case, we were not able to process
multiple hidden layers when we talk
about machine learning. So because of deep learning
we have multiple hidden layers at once now, let us understand this
with an example. So we’ll take an image
which has four pixels. So if you can notice we have
four pixels here among which the top two pixels are bright
that is a black and color whereas bottom two
pixels are white. Now what happens
we’ll divide these pixels and we’ll send these fixes
to each and every node. So for that we need four nodes. So this particular pixel
will go to this node. It will go to this node. This pixel will go to this node. And finally this pixel will go
to this particular know that I’m highlighting
with my cursor. Now what happens we
provide them random ways. So these white lines actually
represent the positive weights and these black lines
represents the negative waves. Now this particular brightness, when we display High brightness
will consider it as negative. Now what happens when you see the next output
of the next hidden layer, it will be provided
with the input with this. So this will provide an input
with positive weight to this particular node. And the second input will come
from this particular node, since both of them are positive. So we’ll get this kind of a note
similarly this one as well. Now when I talk about these two
nodes the first node over here. So this is getting input
from this node as well as from this node. Now over here,
we have a negative weight. So because of that
the value will be negative and we have represented that with black color similarly
over here is well, we’re getting one input from
here which has a negative weight and the another input from here
which again has a negative way. So accordingly we get again
a negative value here. So these two becomes
black in color now, if you notice what will happen
next will provide one input here which will be negative
and a positive weight which will be again negative
and this will be also negative and a positive weight. So that will again come
out to be negative. So that is why we have got
this kind of a structure if you notice this this is nothing
but the inverse of this particular image when I
talk about this node over here, we are getting the negative
value with the The weight which is negative and a negative value
the negative weight which is positive. So we are getting something
which is positive here. Now, obviously, I want this particular image
to get in verse I want these black strips to come up. So what I’ll do I’ll actually
calculate the inverse by providing a negative weight
like this over here. I’ve provided a negative
weight it will come up. So when I provide
a positive weight, so it will stay wherever
it is after that. It’s detect and the output you can see will be
a horizontal image not a solid not a vertical
not a diagonal but a horizontal and after that we are going
to calculate the difference between actual output
on the desired output and we are going to update
the weights accordingly. Now, this is just
an example guys. So guys, this is one example
of deep learning where what happens
we have images here. We provide these raw data
to the first layer to the input layer then what happens these input layers
will determine the patterns of local contrast
orally fixated those partners of local contrast, which means that it
will differentiate on the basis of colors in luminosity
and all those things. So we’ll differentiate
those things and after that in the following What will happen it will determine
the phase features. It will fix
a those face features. So it will form nose eyes ears
all those things then what will happen it will
activate those correct features for the correct phase
or you can say that’ll fix it those features
on the correct phase template. So it will actually
determine the faces here as you can see it over here and then it will be sent
to the output layer now. Basically you can add more
hidden layers to solve more complex problems. For example, if I want to find out
a particular kind of face, for example a face which has large eyes
or which has light complexion so I can do that by adding
more hidden layers and I can increase
the complexity also at the same time if I want to find
which image contains a dog. So for that also, I can have one
more hidden layer. So as and when hidden
layer increases we are able to solve more
and more complex problem. So this is just
a general overview of how a deep Network looks like so we have first patterns of local contrast
in the first layer, then what happens we fixates these patterns
of flow will go Entourage in order to form
the face features such as eyes nose ears Etc. Then we accumulate these
features for the correct phase and then we determine the image. So this is how deep Learning Network
or you can say deep Network. Looks like if you have
any questions any doubt, you can write it down
in your chart box. All right. So we have a question
from our seats again. He’s asking can you please
generalize this a bit fine? I shall tell you. So what happens we have talked
about neurons, right? So you can stack layers
of neurons on top of each other. So you have one year
on then on top of that you can have
another neuron like that. You can stack up or your neurons
or notes on top of each other the lowest layer that is
there will take the raw data. In this case. We are taking images. Although it can be
takes sounds Etc. After that would
happen each neuron or each node store
some information about the data. They encounter after
that each neuron, or that node will
send that information to the next layer of the node, which learns a more abstract
version of the data below since this is obviously
a more abstract version of the data that is
that it is getting right. So the higher you will go up
the more abstract features, you’ll learn this is how a general law
deep neural networks work. I hope you are actually
satisfied with the answer. All right. He says yes. So this was just an introductory
session of deep learning. If you have any other questions
any other doubts, you can ask me basically
we have discussed how deep learning evolved. What were the reasons
why deep learning came into the existence
and what exactly it is. I’ve just given you
a general overview. Justin says amazing session. Thank you. Justin any questions guys. All right fine. So there are no questions. I will move forward and I’ll give you some
applications of deep learning. So here are a few applications
of deep learning. It can be used
in self-driving cars. So you must have heard
about self-driving cars. So what happens it will capture
the images around it. It will process that huge amount
of data and then it will decide what action
should it takes to take left. Right? Should It Stop So accordingly, it will decide what action
should it take and that will reduce
the amount of accidents that happens every year then when we talk about
voice control assistance. I’m pretty sure you
must have heard about Siri all the iPhone users know
about Siri, right? So you can tell Siri
whatever you want to do a little search it for you
and display for you. Then when you talk
about automatic image caption generation, so what happens
in this whatever image that he will Then Gotham is in such a way that will generate
the caption accordingly. So for example, if you have say blue colored eye
so it will display a blue color dye caption
at the bottom of the image. Now when I talk about
automatic machine translation so we can convert English
language into Spanish similarly Spanish the French so basically
automatic machine translation, you can convert one language to another language
with the help of deep learning and these are just
a few examples guys that are many many other
examples of deep learning. It can be used in game
playing it can be used in many other things and let me tell you one very fascinating
thing that I’ve told you in the beginning is
well with the help of deep learning MIT
is trying to predict future. So yeah, I know it is growing
exponentially right now guys. So this is it for today’s session on
what exactly is deep learning. If you have
any questions any doubts, you can write it down
in the chat box. Any questions guys. All right, so we have
no questions will move forward and I’ll just give
you a quick summary of what all things
we have discussed. So first we saw why
we need artificial intelligence. So basically things that humans cannot perform
be want a You need to learn and do those tasks for us. Then we understood what exactly
is artificial intelligence. And what are the various subsets
of artificial intelligence like machine learning
deep learning Etc. So we first focus on machine learning
we saw what exactly it is. We saw various types of machine learning namely
supervised or unsupervised and reinforcement. Then we focus on various
limitations of machine learning that led to the evolution
of deep learning. Then we saw a use case
in which we saw how to recognize an image
using a deep networks. So this was it
for today’s session. This video will be uploaded
into your LMS. You can go through it. If you have
any questions any doubts, you can bring it
in the next class or you can ask our 24/7 support team as well. Thank you, and have a great day. I hope you enjoyed
listening to this video. Please be kind enough to like it and you can comment any
of your doubts and queries and we will reply to them
at the earliest do look out for more videos in a playlist. And subscribe to our Edureka channel
to learn more. Happy learning.