Session Information:

Cloud Computing for Data Analytics Applications

Data Analytics is the foundation of modern decision-making process It allows analyzing raw data tomake important decisions. During the recent Covid-19 virus epidemic we saw the power of analytics topredict the severity and the duration of the epidemic.

In every decision made by hospitals, medical professionals, and government used analytics to plan the response to this disease. For every busines sexecutive it is essential to understand basic principles of analytics. First, this seminar offers basic information about analytics which includes machine learning and deep learning. Next it enumerates the tools necessary for building analytics models. Analytics models are currently built using TensorFlow/Keras, Python/R software products and require asignificant amount of computing resources.

This precludes running analytics models on personal computers. Generally, they are not powerful enough. Analytics models usually run on cloud servers because they have enough computing resources. Consequently, Fortune 500 companies build analytics models on cloud servers. Cloud servers provide access to GPU (Graphics Processing Units) and TPU (Tensor Processing Units).

These processors can execute trillions of floating instructions per second. Due to parallel processor architecture built in GPU and TPU, they are ideal for matrix multiplication operations—the most frequently used operation in analytics models. The 3 most popular cloud services are AWS (Amazon Web Services), GCP (Google Cloud Platform), and Azure from Microsoft. Cloud servers provide general computing service in three “flavors” –Infrastructure, Platform, and Software.

They also provide the services uniquely necessary for analyticsprojects. This seminar will examine these cloud services in detail. A comparison of them will be explored from theperspective of which cloud service is better suited for analytics models. Next, the seminar will focus onhow to set up cloud servers for analytics projects.

Session Transcript:

Hello, everyone. Welcome back to RPA and Intelligent Automation Life. Our next session will be about cloud computing for data analytics, applications, the receipt of some of them. And then at the very end when we have time for Q&A, I'll be probing some of those questions And relaying them back to doctor Ashe. So without further ado, let me introduce our distinguished speaker today for this segment, doctor Ashe Power, an educator author, and entrepreneur and technology visionary with three decades of industry and academic experience. He has founded several successful technology companies during his career. Doctor Power teaches courses at the California Institute of Technology Caltech, Pasadena, and the University of California system, which includes UCLA, UC, San Diego, UC Irvine.

Doctor Power, it's a real pleasure privilege to have you here with us today. And we really look forward to your presentation.

Great. Thank you very much, Jose. Let me start my presentation now.

All right, so I hope everybody can see my presentation. At this time and so here. I will be basically talking about the cloud computing in the data analytics application. I've been hearing the other presentation by other speakers as well. So, they were talking about robotics And the question that comes up is that we're already going to build these applications for robotics or for or for data analytics. So, the answer for all of this thing is that when we do these things on cloud computing and so, my presentation, what the next almost 30 to 35 minutes will be based on what is cloud computing and what we can do about it.

So, here is a very, very brief bio. And I already had already mentioned to me that I'm currently teaching at Caltech beside that, I have also been an entrepreneur. So, now, let us move to and these are my courses at Caltech, if you need to get some information about it. So, now, this is the outline of my talk. First of all, I want to talk about the power of data analytics, to me, how important data analytic have become. And then I'll talk about the cloud computing and the tools that have been used for the data analytics. And then I will also discuss the Google Cloud platform. There are a lot of applications that can be that has already been prebuilt on the GCP. GCP, started for the Google Cloud Platform.

So I'm going to show you two applications, which are sitting there. one is on the vision side, and other one is on the speech to text speech to text application.

OK so now let us look into first of all the Power of Data Analytics. So I'm sure everybody must have experienced this covert 19 pandemic which is going on from the last 3 or 4 months. And that has literally changed our life. And one of the thing that that comes out from this pandemic is that how important it is to analyze the data. So let me show you a few things. I'm bringing my pen out right now. So the importance of data analytic and this thing is very, very far.

That's what basically, we are going to talk about the power of data analytics. So, here, of course, if you look and go to any hospital, any pharmaceutical company, any treatments, or any vaccines, which are coming out, there are all dependent upon big data. The data is the most important thing. And for the last three months, I really appreciated data analytics, quite a rock. And this data is, of course, is being shared at the federal level or at the state level, No matter where it is, or even at the city level, they are using all kinds of data.

I'm sure everybody must have seen this, this kind of a charge these days, which are appearing like this one, how the total number of cases went up, and then hopefully, these things are coming down, all newspapers and all websites are displaying this kind of data, so all this data is being done using data analytics. So, data analytics has become really part of our life.

So, another major application of pose for data analytics is Amazon. And so, here, of course, as you know, that the price of the Amazon stock is going up, and up and up, every day did gain some, or goes up a little bit. And why the thing is happening, why Amazon has become such a successful company because of data analytics, because they analyze the data. They basically predict what the user wants and give them what the user really wants and ship them out those products. So it has become very, very successful, and because of this, also.

Message, Jeff Bezos, who owns this company has almost become the richest man in the whole world resort as your nose. as Bras are more than 100 billion at this time. So, this is, these are just some applications of data analytics. Another major application of data analytics or post is autonomous cars, the car that driving all by themselves.

We face a very sad statistic that a person dies every NaN, so sad. But that is good. Hopefully, if we use ... cars, we can solve this problem.

So, this problem of autonomous cars can be solved using the vision, AI, AI for Vision, Artificial Intelligence for vision. This thing is all a part of data analytics. So, now we have seen three major applications. First of all is the first of all, we saw the covert 90.

Data. Then, we saw Amazon.

And now that we are also looking into autonomous cars, all of these applications are, actually are the data analytics applications. And so the question really, is the center, where can we build these models that can do these, kind of a fantastic things that we are seeing in, again, in real life?

Just want to show you, so, very interesting, Very likely, this is the baby, right?

So that in the future, may come about maybe 10 years or 15 years now, but somewhere around there. So the car would be that you just punch in your particular punch in your GPS, but an edge of your destination, and they've gotta just going to take you there. So, this is the way, the thing that is happening. So, not that many different levels to these cars are, currently, we are, at level three. We will have someone, I don't know most features of our cars. But the future is basically this, which you are looking at it over here. So this is the way the future is, and all these things are happening. Because of data analytics, people via basically, drawing, we have, basically, trying to use data, and see how we can make sure, again.

Now, coming back to that, what is, coming back for the second step here, what is data analytics, Data Analytic Kind of different names? Some people call the senior data mining, predictive analytics, or machine learning, and these days, the word is coming around, which is called deep learning. These are all part of data analytics, and those are the applications that we just saw before, can be derived from the sea things. So if there are many different modeling methods out there. So now I'm gonna get into the technology part of it. So, there are total, 11 modeling methods, and so these modeling better, of course, regression, logistic regression, discontinuity analysis, K nearest neighbor decision at admission fees, Naive Bayes Neural Network, Blustering, Principal Component Analysis, Support Vector Machine. And so our focus will be primarily, or the neural net, but these things have become very, very popular these days.

And deep learning basically uses use neither. There are other modeling methods as well, but our focus should be on the neural networks. And these applications, also, for the vision that we just saw, just a few minutes back, for autonomous cars, can be achieved on the neural networks. And to build these models? Or will we have to go to the cloud?

So that's what the focus is that and where can we build these modeling these kind of a Margaret's. So, now, let us also look into what is Artificial Intelligence and, and also, what is machine learning. Artificial Intelligence is a much bigger area. Machine learning is the part of artificial intelligence and deep learning, which is becoming very popular these days. Deep learning, as also becoming a part of machine learning. So now, my focus would be that, how we can use Cloud.

To basically build a mile deep learning, and deep learning models. So the ones that are being used for the vision, that we give me and machine can able to identify all the objects, which are in Africa, in an image. And based on that, only, it can drive the car once, once a car knows that, what are the objects right next to it, where the pedestrians and various other cars coming in, and what is the traffic light? And all these things can be recognized by the car, so the car can drive all boats.

So now let us come to cloud computing, which is the focus of my talk. So what is cloud computing? So now we're going to basically cover in more detail, what is cloud computing and who are the vendors in this in this industry?

So let's try to understand what what the cloud computing is. A very brief history of cloud computing from 19 70 onwards, I did all that thing. Most of, the big businesses, I used to have a huge computers in the 19 sixties and seventies, most of the big business used to have IBM 363 70.

kind of a mainframe computers. Then, after that, many, many, new, mainframe computers came about, and that is what we use in most of the MSP. And so, the problems with this approach was that, that the business has to spend a lot of capital because these machines are not cheap. They used was, I don't know, five to $10 million each. And so there was a lot of capital investment. And also, not suppose, if, you load increases more, number of people who want to do business with you, then you cannot change. The configuration is inflexible?

And also, they used to be when I recall that when I got my first job, they use a part of a whole team. And then most of the time business is used to have between 50 to 100 people kind of maintaining these monsters over there. So there's a lot of, well, the fact there was a lot of expense, to basically use my resources for all of them. So all of this has been changed by cloud computing to another cloud. Computing has come about in early two thousands. And basically, what is cloud computing?

Cloud computing basically has your computers, a very it and store data. And all of them are available on the web.

The Cloud computing is to use the remote servers on the Internet to store data.

So what is the advantage of a cloud? It's a win-win situation. In fact, this is very, very important thing. because for businesses, they don't have to spend money on capital, on people to maintain their computers. They basically hire computers from companies like the Google Cloud Platform or of Amazon Web Services, so, and also, this is the, become a very, very profitable business for Amazon as well. So for Amazon, as you know, HUD has the Amazon Web Services and also.

So there is this thing have become very, very profitable for you, for Amazon to run these cloud computing.

So in so this is a win-win situation business basically can save a lot of money and these guys can make a lot of money as well.

So what are the benefits of cloud computing? The benefits of cloud computing are eliminated. That capital expenses, you do not have to spend a few million dollars to buy these computers, but you can get it rented these computers from the people so you can save a lot of money here and only pay for what you use. This is very cost effective, and performance is always better because these people from where you are renting these computer, which is AWS, or it could be GCP, they basically maintain the software for you.

So, there are no virus is coming out, you don't have to worry about other things, everything the beta updated, because these kind of our servers are being maintained by professionals so they know exactly when to update your software up, the newer version of the operating system, and remove any bugs and all those things, And and the productivity improvement in there. Because the IEP team, people who are working at the businesses not basically focus on the customers, is very, very important for the businesses to use these services, because they don't have to hire people and those people who are already there. Yeah, basically, working with the customer that this time, so this is a win-win situation if you have a cloud.

Another major advantage for a cloud computing opposes, scalable architecture. You can If you're loading reason, you can add more processors, suppose you are basically have one processor on the on the cloud. And here you are paying a certain amount of dollars. But suddenly your load ingrained in, you can say, OK, I'm going to warfarin servers. I will keep on really looking. So you're right. The role of temporarily, but above the load is not there, then you can take the scene saw from your, from your data, from your architecture of computers. So now, you will, of course, going to pay less. So this is a scalable architecture, is very flexible, and also, you have better security of good data.

Although we have seen some cases when, even in the cloud, that there is going to be a security to the beach has other, but, but. But if you have data on the cloud, it is usually more secure than if you have it on your own. If you have, if you maintain your own computer, then it's really not made it so somebody can really break in and steal your data, although, in a cloud, security is not 100%, but still it is pretty, pretty good. So your data basically remain secure, also in the cloud computing. And cloud, of course, makes life simple, or the businesses, as well as for the vendors.

Now, so now, if you have an Analytics job, because that's what our focus is, that, to how to do that thing, you should not run your analytics monitors on personal computers, because usually, they are not fast enough. We need a lot more CPU power. And we want to make use of GPUs. And also, we're going to make use of TPUs, which are available on Cloud GPU. Stands for the Graphics Processing Unit, and Tensor Processing.

So this is the way to go. And this, of course, the answer for Crude Run Analytics model, is that we have to go to the cloud.

So now, you can rally of cloud. Thing in, many different, when you have your architecture, you can run this thing and Infrastructure as a service or platform as a service or software. That's, most of the time, people use this thing in the company. The hardware is being provided by you, by the vendors are the single being provided by AWS. You decide your own operating system, you would say all of your software and values around your applications on there and the hardware is going to work flawlessly without any interruption and that is the responsibility of the vendor. So most of the time, people use these things, you can also use platform as a service, that even the operating system and all the other stuff, can be decided by the vendor. Or you can run software as a service also, As you know that that.

Google also has a service called Sheets, which is just another way by which you can do the spreadsheet. And also slides, which is just a replacement of the PowerPoint. So these two things are basically running in cloud, which is software as a service. So you have three different variables in which you can use the cloud, which you can use the cloud.

OK, cool Autumn! They have ended up into the mega vendors here. So for the three of them, wanted AWS, which is very, very famous and also very, very profitable, they made tons and tons and tons of money with it.

And then another one, of course, is the Google Cloud Platform.

And the Microsoft ..., these are the major players, but if you look into the history of it now, via since it isn't a profitable business, cloud computing. Now, of course, everybody wants to jump in, it might be what you want to make some money on it, but this is a family history of it. A W S started in 2006 roughly. Then of course, in the Windows, Azure and infant GCP, which is the Google Cloud Platform, around 2013 onwards. Now, after that, of course, everybody wants to get, basically jump into this area: Now, IBM Watson has an AT&T as it arrives, it has it, and VM Ware, Cisco, Oracle, and so on and on, and tons. and tons of people out there. And again, this is not a complete list. Them feel more of them as well.

But this shows that the basically who are the major players in cloud, so cloud computing is dominated by basically by Amazon and also by Google, Amazon bigger share. But since we are using some Google products, it makes sense for analytics jobs to basically run on the Google Cloud Platform.

So now, let us look into the tools, which are available. And what are the tools we're going to need to build a minority job? Why our goal is to do data analytics? So this is what our, our primary application is. So let's see what other tools you're going to use. So on the hardware side, of course, we're going to use cloud, we can use GCP, this is Google. You can use AWS, Amazon Web Services, or Azure, or any other one. There are many, many, other one as well. And these things now they have availability of the GPUs and TPUs, which are absolutely necessary to build your monitor, that on the software side, we're going to use TensorFlow and Keras. These are the new pieces of software, which is coming up from Google, and they allow us to build these models. So here is our TensorFlow that you can use, TensorFlow and Keras.

They basically go hand in hand, and terrorist is just an interface software.

So TensorFlow is a major piece of software, and we certainly not TensorFlow to point, or has also been released earlier this year, which is now much, much better than TensorFlow one point O. And so this is the way to go. And TensorFlow 2.2 includes kara's.

So another tool that you can use this with just for the Colab. Colab is basically in the platform on which you can write your Python code. And you really know your Python code to basically build your model. So this thing, or the other release, two years back. And co, lab knowledge is being used standard platform. So again, Colab also runs on the cloud.

That's the beauty of it. And so it's part of our lab of some people. Earlier, it was called for laboratory. But or laboratory, at least, it seems like a pretty long word to say. So they made it into a shortfall, and just call that thing is a boiler. So, it is a Jupyter Notebook Environment. There's no setup.

Now, the beauty of all, I believe that it's free, but that's the best part of it.

So you have a whole app, here, and all lab is going to interface with GCP.

Not a problem, with the poll, and so forth. All app is free.

But this is not free. It's not free. This is the place that wants to make some money. But on the outside of what they do, the free. And you add, the also free GPU access also, and support Python two and Python three and, and Colab is integrated with Google Drive, where you can So this acts like a hard drive or for Paula.

So why use Colab? So now why do GCP? No other questions really decided, why should we use GCP? Because you plan to use TensorFlow, you plan to use ..., GPUs, and TPUs. So these are all all Google's products. So that is the reason why you make use GCP. And again, I'm not promoting GCP at this time, but you also have a choice of building these models also on Amazon Web Services, as well. But Amazon Web Services does not you can run TensorFlow on the on Amazon, as well. And you can also run terrorists are also on AWS. But just because TensorFlow and terrorists are the Google's product, it makes more sense to run these seeds on Google. Although the pricing of GCP is the lowest at this time compared to the US.

Sorry about that. I just circle to turn off my phone. OK, now, I know why GCP. OK, now, if you plan to use the CNN, but you use for vision application, then you must use GCP, or, sorry, the GPUs and TPUs. You cannot build this model on your own personal computers, So if you have your CNN.

And then, of course, you can build a CNN on GCP with a GPU. So this is another reason why we should use GCP to build our analytic models.

So now, let us look into GCP a little bit more closely. Now, the type of the products do they have. In fact, GCP perfectly provides many, many products. So you can go to this place, console dot cloud dot google dot com, and you have an account, if you have a Gmail account, then you also have an account with the GCP as well. And if you stop using it first time, then they also wanted to give you a better for $300. So you'll get a credit if you use GCP.

And you can start building models, and you see, OK, so they also provide other products, also, But I think the focus of my talk today, is the only data analytics will show you those features of the of GCP that can do with data analytics. We can, basically, a machine learning and also, deep learning models, which are available on GCP. But GCP also provides some field service, as you can, build your own computer, that you can also have a lot of storage and networking and many things out. So, but that is not the focus of the talk today. My talk is primarily on data analytics, on what to focus on, on the type of pictures, which are available on the, on GCP. So these are the five different API, which are available on GCP. First one is Vision.

As we saw the application for Vision, this is really phenomenal. Light. You can, you can build application. I'm wanted to show you, also, an example of vision, as well, and which can basically analyze your image. This is very, very important, and really mind boggling, the applications they have. You can also have natural language, you can even do the borrowing of international language and find out the sentiment, you can also do the speech from speech to text. This is really, really, basically amazing. You can feed into your audio into text. So also speech, basically, is that you can convert audio into text. Then you can translate.

This is something really, really amazing, that you should try it out, and basically says that if you have some document in English, it will convert that into Spanish, Or it can also learn about the same language, same English, text into Japanese, an, all, or any other language. There are 100 languages that are being supported. So this is something really amazing, that this is the prime application for artificial intelligence. If you wanna see AI, then use this application, but this is just mind boggling. Andy's Margaret is almost ready on GCP. You can use. And the last one, of course, is a video application, which can analyze your video and see what kind of a images do you have in your video. So all of these things are GCP, which is called machine learning, whether you call it is seen as a machine learning, or whether you call this thing as a deep learning. It is the same thing, but basically here, these models are already available on the Google Cloud Platform.

And basically, you can start using these models, OK. So this is basically the work. You can write your program and Paula or you can write your personal computer or you can have a website from which you can use or you can even run these Panama martyrs on mobile phones. And these are the five APIs which are available at this time: vision, natural language, speech, translate, and video. Now, in order for you to do the authentication, if you're the owner of the account of GCP rates, this service account key. So this is the key file, which is nothing but a JSON format.

And this, you have to generate yourself, are the owner of this account, will, will do it. And this, E, get attached with your application.

That is the only way under GCP or Google Cloud Platform, will know that this is a valid application to run. And so this is the way the security system with equally, that is being built into GCP services.

So now I'm going to show you an application, but another maybe 3 or 4 minutes left. So I'm going to show you an application in a picture of me sitting with my daughter, and this is displays in Italy. And this is Venice, as you know, in wellness, we have bundled up votes. So I'm gonna analyze this image. So I feel the same way and Here we go.

First of all, the system is going to say, there's a Fifth and it says that, Hey, there is a joy here. I want a Phase one. And Phase two also Have not shown the competing So it can recognize, OK, this is a joy, it. There is no sorrow, sorrow is very unlikely. And there's no anger in it.

I'm sitting with my daughter here, sort of ..., no surprise, no exposed nobler, and there is also some hardware. And I guess my daughter is wearing something of the head, that's really the NSW.

Then you can see what kind of objects you have looked at this. All these bounding boxes. Are we created on these images? and you feel that you have a law says, yes, I'm wearing eyeglasses. It is of course, it is out of air. Outerwear means I'm wearing a jacket here, and this is a person, and also Luggage. And that back, as you know, my daughter, as a person who had, an person, wasn't getting up, and, and all of these seem to be recognized by, by, by the AI system. So again, here, basically, what I'm using is, is my GCP vision services.

And now, once we have the ability to identify the objects in an image, this type of Vision API is also used in autonomous car. This is the same thing, we saw some application for our ..., so the same features are same same technology that have been used to identify these objects. In an image is, is used also in a drama scars as well. Then also you can define your labels as well. So now the devil's's eyes even smile iris tourism. Sending vacation. It says that these are the type of the votes that can be used to describe the symptoms.

And I'm not showing you other features as well. It can also do OCR it and deliver text. It needs than any kind of our text, which is there in an image, can be converted into real text the text in an image, using the OCR Technology Optical Character Recognition can convert into real text. So, I'm not showing you that showing you. the demo of that, because I'm limited by only 35 minutes, Daniel. You also have, will show some properties, and it can also recognize if this content is added oriented or not, so those things will also Florida.

Now let us look into speech to text. So it says, I'm again going to going to give you an example of speech to text. So and also you can even go to speech to Text. That is one of the API that is available. And here you can go to cloud dot google dot com slash speech to text. You bet even take one minute of audio of your own, and this is a, you can test it out and see if you can convert your speech into text as well. OK, so here is the speech, I'm not going to play my audio here. But this, this is a human constellation of my audio. I did this constellation and that is what my audio is. So I did this year, this conversation and look at that translation that is being given to me by the system. As you can see that this transition is very, very similar to the actual translation of my speech. So this is a mind boggling application.

Yeah, you can take any audio, and then word this audio than word with audio, into text. And this is really remarkable because I had a lot of different applications. Basically, you want to type up something rather than trying to type something up. You can speak if I speak something up and get converted into text. And the text is very, very good at all. Until May there are very, very few error, So that if you want to generate some text, and, you know, it is difficult to tight, but it's a very easy to talk it out. So here, you can speak something up and convert that into text.

Excuse me. Excuse me, for a second.

OK, so that is the end of my talk. I'm sorry, that was some, that interruption, of course, because of the technical problems. But anyway, let me summarize. Our motion is 9 36 right now. So I'm just going to wrap this thing up. And so first of all, we saw the power of data analytics, that how data analytics. So important data analytics of not only deals with the data, but also deals with audio, and video is up.

And then, of course, we have this cloud computing, how to solve these problems with cloud computing. We saw that boost for cloud computing because TensorFlow and Keras, we looked into GCP lot more detail. and they saw the AI applications, vision, and speech. Another application, which is worth looking into is that translate? The cab, not shown you before the bladder lack of time, that here you can take some English language stuff, you'll have them documented in English. And it can convert that into whichever language you want. Bill, you've been converted into Spanish, or any of the hundreds of languages available. So this is the way to go. And that is the future of data analytics that we can use, artificial intelligence, foods.

So that's all I have to say. Thank you, very much. And now, we'll open it up if you don't answer.

Doctor Ashe, thank you very much. You may now stop sharing your presentation so that the audience can see both of us on the screen. All right.

You can keep your camera on to stop sharing your presentation. Yeah, OK. Here we go. You got it. You got it. Thank you. Thank you very much. Wow, Wow. That was quite a journey on the evolution of cloud computing and the and the and all that's going on up to today. We had a number of questions that came in during your talk. So I'm going to relay those to you, and the, for the audience, if you are, as you still have time, go ahead and answer additional questions. And I'm going to be picking over those questions and letting doctor power know about it. So the first one is that, of course, from the very beginning of cloud computing, a lot of the concern of cloud computing had to do with security and the risk associated with the how secure it is to have your security and data integrity on cloud systems. And we have gone a long way with that.

And just just like to have your perspectives on where we are today regarding security and integrity with cloud systems.

Great weather. Good question. So let me let me advance the signature. Security is very, very important. And all the cloud is not 100% secure. Way to start with, I should tell you that it is not, but it is a very, very close to perfect security. Should we have seen a lot of examples where the data has been stolen from cloud as well, but you need the security that is being provided by the cloud is usually better than the security that we are going to enforce on our own computers. Our computers are also vulnerable. As long as your computer is connected to the internet, somebody can hack into it and steal your data. So, compare to the security that is available on our own personal computer, Clouds provides much, much better security, although it is still not 100% secure.

Very good, Very good. That's a, that's a great insight. Now, Michael Levine had a question about the, the trends in cloud computing. If you look at what has happened and then just in the last year alone, what do you see as the most significant trend in cloud computing cloud computing in the past year?

The most significant improvement in the cloud computing is that now we have these artificial intelligence models available for us to use. These are pre trained models. Earlier, like if you go back to 2016, 2017, we can still use cloud. But we are using cloud primarily for a data about creating a computer on networking, and all those things. But from the last 2 to 3 years, we have no in-store, lot of pre trained model. So I can do my vision analysis, video analysis, audio analysis. And all those things on cloud use models for not there before, and these models are getting better and better day-by-day, because more data you have to train these models are models are going to become better.

And the same kind of services are being provided by AWS as well as GCP. But since we use TensorFlow and Keras to build these models, it makes more sense to use GCP, which is Google Cloud Platform. Although the same thing will also want to work on AWS as well.

Very good.

And you talked a bit about the evolution of AWS and the and then how the Google Cloud Platform has has come to into prominence maybe in the last year or two. Here is about AWS capability. Is that, does AWS also provide artificial intelligence services like GCP does?

Yes, they do. And so, on AWS is trying to compete with GCP. These are the two primarily cloud services that are available that can do artificial intelligence. They have tons and tons of models over there. And you can also run you can also run TensorFlow and Keras on AWS as well.

So AWS and GCP are competing very heavily and with each other. So most of these models available throughout the whole, in both of these platforms, but my preference for GCP is that because we are using TensorFlow and also get us to build these models. And these are the Google's properties. So usually these, these pieces of software usually work very well with the tea with GCP, And also, we need the GPUs and TPUs, Tensor Tensor Processing Unit. And this is a piece of hardware that has been developed by Google to basically solve the TensorFlow problems. And this thing is not available with AWS. So again, these are very competitive products to GCP, as well as the blue eyes. And in some respects, GDPR, the edge or AWS. But AWS is also a very, very good cloud service provider.

On the, on the GCP portion of your presentation, you had something about the GCP Vision API. And in GCP, Video API, what is the difference between the vision and the Video APIs?

Great. So innovation, we can analyze a single image, we can, as I showed you, the example of my own image with them where I'm sitting in an image with my daughter in Venice, Italy. But, so, it can analyze different features of the image that, what, the objects in the image? That is? The Video API basically deals with the complete video. I suppose, if you have one hour of video, and let us assume that you have only one, like, a NaN of a baseball going on in the complete one hour of video. It can identify that, Hey, there is some baseball, also in this video. I suppose if you have five hours of video, and you need to analyze that, someone has to watch five hours of video, .... But in the company does can read this entire video file, and analyze it, what kind of objects you have in it.

So, videos basically worked on the video segments, which are used a lot, whereas the Vision API is primarily proposed towards a single image.

Yeah, it's fascinating to see the cloud computing is at a stage today, where you can have a single entrepreneur, with access, with processing power, and capabilities that maybe a decade ago, were only available to multinational corporations and governments. And now, you can have entrepreneurs, really, literally, anywhere in the world, who can access that type of capability. I'll ask one final question related to these pre trained artificial intelligence models that you talked about, are there available on GCP? Are they are they sat, or you can train them further to get better results?

So we already have some pre trained models that can be used by anybody in public, and they usually gives a pretty good results, But you can make them even better. For example, if you want to train these models for medical images, you can not take these pre trained model and not train this model further, by adding not thousands and thousands of medical images, Now, this marker, which will be created, will become even better for your specialized images. So again, there are a lot of a lot of images out there, which is in the medical industry, also in the sports industry, and in a whole lot of other areas, security industry, There are tons, And tons of videos are there. So these models can we can further, for your own specific applications. And for that, you have to use the services and write your own code, and play them also further. So Google has provided some basic tools, which are going to work for most of the images, most of the audio, most of the video.

But these models can be made even better with your own videos on own images.

Doctor Ashe, thank you so much for your time here with us today. It's not. It's not every day that we have an opportunity to have, not only a world-class educator, but also an entrepreneur who shares this level of the practical knowledge with all of us. So, thank you very much for being with us today.

Sure. You're welcome. Thank you.

Thank you.

Bye, bye.

Ladies and gentlemen, we are concluding the presentation for today. And tomorrow, we have an exciting they lined up for us. Today, we had wonderful presentations from experts in the field from educators and entrepreneurs like doctor Ashe and tomorrow we'll have another set of amazing presentations. We'll start the day at the same time as we started today with Wesley, ropes, and from the VP of Research and Development and Technology Transformation from Kroger. And he's going to talk to us about the transformation, to intelligent enterprise, going to focus on cultural aspects, business transformation, aspects, and digital transformation aspects of creating a true intelligent enterprise.

That will be followed by Margaret Moore, who is the Director of Quality and Transformation, and Morningstar and he'll give us that financial services perspective on, on the title of his presentation is, Control Your Robots, Designing an Effective RPA Governance Program. So, we're going to talk about governance. There are a few questions. So they relate related to socks and financial processes. You do not want to miss a session that's going to focus on that, on that topic with IBM practical applications in. In one of the industry, leaders will have, then, Brandon, not, who is a Senior Vice President, A UI path given us, a, talk on a robot for every person. And a bit of the history of the industry, and where the where we are today, and where we're heading, and a real holistic approach to digital transformation that brings humans and technology together. Seamlessly, you do not want to miss this session. And we're going to conclude tomorrow with a fantastic speaker. ...

is the vice-president for Enterprise Project Management and the Strategic Initiatives at Caesars Entertainment. So, he's going to be talking to us directly from Las Vegas and where we're going to learn about how robots are being used and casinos and entertained the entertainment industry. And, and really, this is a phenomenal speaker. He has contributed over a billion dollars in ... on the, all the improvement and innovation projects that he has led for, for a number of large organizations. So, exciting day tomorrow. I appreciate every one of you spending time with us today.

Great sessions today. We're going to continue the conversation on LinkedIn So if you look at the link that I provided during our session. You can ask questions there, you can make comments on there. If I receive some request to connect directly with me, that's my name. A LinkedIn shows that. Paris, I am over my 30,000 connections limit. But if you, if you connect with, If you try to connect with me, or if you follow me, I'll follow you back, and then you should be able to to establish a connection that way. But thank you so much for everyone who participated today as you close the session. Again, there will be a popup box that they'll give you a survey that you can provide feedback on today. And I look forward to seeing you tomorrow.

Have a good day.