Sudheer Matta, VP Products, Juniper Networks

The ABCs of AI

Sudheer Matta Headshot
Image shows a small photograph of Sudheer Matta, VP Products, Juniper Networks on the left, and a slide title that says, “Examples.” A large blue circle on the left has text that says, “Artificial Intelligence,” and a line goes from the circle to an Amazon Alexa logo.

Cut through the AI noise: How AI can help IT.

There's a lot of buzz around AI and how it can help IT operations. There's also a lot of confusion. You’ll discover the difference between AI, machine learning, deep learning, data science, and other common techniques in this video.

Show more

You’ll learn

  • Real world examples of AI, machine learning, deep learning and data science 

  • What Juniper’s Matta says is the biggest movement in AI in our generation

  • How all of this applies to us – and why it all matters

Who is this for?

Business Leaders Network Professionals


Sudheer Matta Headshot
Sudheer Matta
VP Products, Juniper Networks  


0:00 my name is Sudhir Mata I'm the VP of

0:03 products here at mist today we're gonna

0:05 talk about the ABCs of AI what we will

0:09 do is actually cover what is AI what is

0:12 machine learning what is deep learning

0:14 what is data science without further ado

0:16 let's get into it so AI machine learning

0:21 deep learning data science are very

0:24 interesting different terms and and they

0:27 all represent different things and so we

0:30 first set a level set on the baseline of

0:33 what these definitions are artificial

0:36 intelligence right this is actually the

0:39 the big umbrella of what we are living

0:42 in today our worlds are being

0:43 transformed with AI so the best test of

0:48 AI is is the Turing test that was done

0:51 but proposed by you know Alan Turing

0:55 Professor Alan Turing you know five six

0:58 decades ago right what is the Turing

1:00 test the Turing test is quite simple if

1:03 you as a human are interacting with a

1:07 system a and a system B and if system a

1:10 behind the scenes is a machine and

1:13 system B behind the scenes is a human

1:16 and you as a user can tell the

1:19 difference that's AI you have arrived

1:21 right so so that's really at a very

1:27 macro level AI is something that passes

1:30 the Turing test and and I think you know

1:33 slowly but surely in networking we're

1:36 going to get there right

1:37 so this is our crusade and we're going

1:39 to show you some examples of how we're

1:40 getting there number one next what is

1:43 machine learning machine learning is a

1:46 set of models and an algorithms to help

1:50 you get to AI right and so there's all

1:53 kinds of models here that you know very

1:58 basic you know regression schema

2:00 regression models you know to really

2:02 complex models out there and machine

2:05 learning is collecting a lot of data and

2:07 actually and if you can train the

2:10 machine to learn from the data

2:13 that's that's called supervised machine

2:15 learning which is what Google did many

2:17 moons ago when they were trying to

2:19 identify the picture of a cat they they

2:21 fed you know millions of pictures of

2:24 cats to the machine and then when the

2:27 very next picture was was was input into

2:29 the system it was identified as a cat

2:31 right and so that's that's supervised

2:34 training the base training based machine

2:36 learning the next one is unsupervised

2:40 machine learning which is one of the

2:41 principles we apply here in mist for our

2:44 location engine when you take your

2:47 iPhone and you broke walk into a

2:49 hospital you know or then are you

2:52 walking to an airport or you walk into a

2:54 stadium the how the iPhones are of

2:57 characteristics behaved is wildly

3:00 different so V machine learn on the same

3:03 phone in different environments without

3:06 you pre feeding hey this is how a

3:08 stadium looks like and this is how a a

3:10 hospital looks like or whatever right

3:12 and so that's unsupervised machine

3:15 learning so we're gonna talk about both

3:17 of those what's deep learning deep

3:20 learning is is the human science trying

3:24 to emulate the human brain right our

3:29 brain is is is millions and billions of

3:32 neurons connecting and and so there are

3:36 layers and layers of neurons that are

3:39 connecting and so either eliminating

3:42 pictures either is or forming pictures

3:45 as we are looking at things and if you

3:48 can think of it today there isn't a

3:49 camera that emulates the human eye sight

3:51 and and and there isn't a machine that

3:54 emulates the human brain in terms of

3:55 what we can process around us right and

3:57 so that's a neural network that's a very

4:00 best neural network in the world and

4:02 neural networking and deep learning is

4:04 about you know you know layers and

4:07 layers of algorithms and reinforced

4:09 learning learning from the data and and

4:12 stuff like that right so this is this is

4:14 really good stuff this is the advanced

4:17 machine learning stuff which is deep

4:18 learning what is data science data

4:22 science is basically a little bit of AI

4:25 a little bit of machine learning

4:27 learning but it's really putting data

4:29 together it's predictions it's

4:31 forecasting it's a lot of that you know

4:35 using data to to do to provide guidance

4:39 is data science right so we we have you

4:43 know each of the facets of these things

4:46 coming up into networking and so we're

4:48 good today we're gonna talk about first

4:50 let's go and look at some of the

4:52 examples we have in our daily lives off

4:55 of each of these principles here so

4:59 what's an example of AI in our in our

5:02 daily life I don't know about you but

5:04 you know at my home I have a three Alexa

5:08 devices the the two little ones and then

5:12 an actual big Alexa Amazon echo and

5:16 that's basically it's it's trying to

5:20 emulate a virtual assistant right a

5:22 virtual assistant all of it available so

5:24 you can ask it almost anything of course

5:27 it doesn't know a lot of stuff and so

5:30 that that involves a little bit of

5:31 training in the background and whatever

5:32 but Alexa represents probably the best

5:36 are the easiest to understand AI for us

5:39 at a macro level probably the biggest AI

5:42 movement in our generation is what IBM

5:46 did with Watson right today when you go

5:49 into a doctor in many many large

5:52 healthcare institutions IBM Watson he's

5:56 basically saying you know what hey you

5:58 know Sudhir is lived in California for

6:00 20 years and and so he has these

6:02 symptoms so I think we can you know

6:04 roughly say this medication will be

6:06 helpful for him but Watson is saying

6:08 wait wait timeout

6:09 he's from Indian descent and lived in

6:12 Iowa and then so and and you know of a

6:14 certain age and so when you put all of

6:16 these other things together that you

6:17 don't learn in a textbook maybe I will

6:20 change the medication or or change the

6:22 dosage right Watson is actively helping

6:25 people make decisions and that's you

6:28 know emulating the expertise of a deep

6:31 doctor right so next the examples for

6:35 machine learning for us are many fold

6:38 right the

6:41 an example for machine learning sorry

6:42 about that the examples for machine

6:44 learning for us is nest right so I have

6:47 a nest thermostat at home you know if

6:50 you can imagine this thermostats have

6:53 not been reinvented for a hundred years

6:55 and suddenly you know nest comes around

6:58 and boom there's an innovation and there

7:00 you know nest knows when when I come

7:03 into the home and I leave the home and

7:05 it's basically trying to you know it

7:06 goes into an eco mode that actually

7:09 automatically adjusts the thermostats

7:12 and and and stuff based on people being

7:15 there it's learning it's learning when

7:16 I'm there it's learning when I'm not

7:18 there

7:18 deep learning is what Tesla and

7:23 self-driving networks are based on right

7:25 deep learning is what self-driving cars

7:28 and self-driving networks are going to

7:30 be dependent on and this is layers and

7:34 layers of neural networks that we can

7:36 use to to learn and there's so many

7:42 facets to deep learning and and the AI

7:45 that's driving you know self-driving

7:48 cars it's amazing right so there's

7:50 that's that's right there in front of us

7:52 now

7:53 data science would be me trying to

7:55 predict you know with the Cavaliers will

7:57 be the NBA Finals this is a an example

8:01 for a few more from a few months ago but

8:04 you know it was fascinating you know

8:07 once once Golden State won the first

8:11 several games it became obvious that

8:13 yeah you know there was no chance in

8:15 hell if they if they were up three and

8:18 one that the Cavaliers would actually

8:19 come back and do some damage right you

8:21 know history tells us statistics tell us

8:24 that teams that are up you know by a

8:27 certain margin at a certain point in the

8:28 series are going to win right so so data

8:31 science is just using analytics and data

8:33 to predict some things that is

8:36 statistical now all this is fine and

8:38 dandy why are you wasting an hour with

8:40 us for if all of this is is generally

8:43 available how does this apply to us and

8:45 why does this matter why it matters is

8:48 very foundational to why if you're going

8:50 to spend the next 30 minutes with us

8:52 here why you should do that right here's

8:54 what's here's the proof

8:55 in the wireless industry the number of

8:57 users devices applications and bandwidth

9:02 are growing exponentially I can honestly

9:05 tell you a hundred percent of you will

9:08 say yep that's happening in my

9:10 enterprise if that's happening there is

9:13 a corollary to that the number of user

9:15 complains the number of people saying

9:17 the Wi-Fi sucks is also growing with the

9:21 number of devices users applications at

9:23 bandwidth but there is one thing that's

9:26 not exponentially growing the one thing

9:29 that's not exponentially growing is the

9:31 number of people on your team the IT

9:34 team isn't isn't exponentially growing

9:36 so how does a team that is probably

9:40 grown ten percent in the last five years

9:42 deal with an exponential network an

9:45 exponential demand an exponential

9:47 support volume or if nothing else in

9:51 some cases the teams are shrinking right

9:53 so how do you bridge that gap there are

9:55 two words for this AI and automation if

9:59 you are not doing these two things you

10:02 are in existence with crisis you're

10:04 going to get replaced you're gonna

10:05 they're gonna find a new team these are

10:07 foundational or fundamental for you to

10:10 incorporate into your system you have to

10:12 use AI you have to use automation and

10:15 that's the only way I think you'll

10:16 you'll you or your team will scale very

10:19 critical obviously independent of the

10:24 size of the network you're running thank

10:25 you very much for joining the ABCs of AI

10:28 we really appreciate and value your time

10:30 and your feedback if you have more

10:32 questions or more comments please send

10:34 them to us on our website

Show more