khiT2mPtgPc

hi I do it there you're joined myself
James archery
Ivan Ivan cloud advisor and we're joined
with Jen we've actually all there I'm 13
years old
exactly Jamie what are we going to be
doing today all right let's begin no
dissent for today I wanna onions
Facebook live events we're going to be
talking about how you can co-pilots a
parrot ar.drone to point out with watson
conversation let's begin the showing so
essentially the entire idea behind the
project is that you'll be able to talk
to watson conversation and essentially
kind of chat bot and you'll be able to
some he copilot it I don't have a drone
you can walk in conversation so unless
we can tell all right explain - it works
I'd like to show everybody a really
quick demo of how it works so let's take
a quick look at the drone as we get the
drone into a flight position you can see
this is our drone over here parrot
ar.drone 2.0 okay and on the screen
behind me you should even see that I
have a no trance application which was
able to communicate with Watson
conversation and is able to ask me for
input so as you can see this node.js
applications have started a new
conversation with IBM Watson and Watson
says hello there my name is Watson your
conversational German pilot what would
you like to do
all right now that Walken's asked me
what I'd like to do I can only ask for
anything
now essentially with this project you
can firmly ask for battery information
altitude information you can tell it to
take off you can tell it to land you can
tell it to rotate clockwise or counter
clockwise and you can even tell if you
don't
up or down all right so let's start with
asking it about battery information so I
begin here as you can see in which ask
it for battery and it says I'm sorry if
your drone didn't provide me with this
information can so sometimes when the
story is connected
the special device that I have here it
just says sorry I wasn't able to get a
confirmation yet so now let me actually
just connect it to that device really
quickly and then while Tammy works on
the technical difficulties as we know
software always likes to be shy when
we're demoing one of the cool things
about Watson technology that we're using
with IOT devices like drones is we have
a huge large partner actually based on
the Germany that is using Watson visual
recognition software with their drones
to identify Russ and because this feel
and its be composing if they need to
send people actually out to the site
therefore they're going to save time and
money money and energy and potential
safety from the these people having the
kind of tower on cells that might not be
worth needing to be searched for so
other examples are people are using
these drones again with visual
recognition software they're training it
into oil pipelines so they can see what
are the insides of the pipelines like
order and you can't have people go in
there so it's nice to have that aspect
to check it out and get a better closer
inside view on that special capabilities
another example of what we're using this
capability again with visual recognition
and controlling the drone in the
conversational aspect is out in
California or using the Watson
capability to identify drought areas
with the drones flying over and
identifying which people are actually
violating for lack of a better word the
regulations in around whether or not
that can be using the sprinkler systems
all day every day right so you can see
there's a lot of applications for using
just even just for the drones Watson's
capabilities on top of maybe leveraging
it or as you've seen or heard with how
Amazon is going to be using them
potentially to deliver packages however
is possibly thinking about using them
for delivering food right even maybe now
medical supplies may be your emergency
situations where people are in a
distraught and if you want to do for
emergency rescue services right and can
use those drones to identify issues so
what I'm saying are we good
I keep going all day every day on this
and that's how awesome IBM is and how
often we are available to work with you
so one of the cool things that Tammy is
going to show it
this is publicly available on the
Internet so you go to bluemix.net and
type in Watson these capabilities are
available to you know continuing though
let's actually take a look at how
Watson's able to control our drill so
now let's begin so so as you can see if
I ask it for battery information it says
that the battery percentage in your
drone is currently 100 percent aha the
Raspberry Pi shut off by mistake so now
it's back around and our server is set
up again now let's say I'll continue
like you of course also asset for
altitude information so now we see them
see I asked it for altitude information
and since the drum is lying flat on the
ground
is that your drones current altitude is
zero if you'd like you can tell me to
increase or decrease the altitude alpha
drone along with a rate of descent or
event from 1 to 10 all right for now
let's say we wanted to send commands to
the drum now these commands could be for
kicking off landing rotating going up or
down all these of the commands so let's
try it out now one more thing I just
like to say is that this drone is quite
loud when it is mid-flight so I won't be
able to talk to you while the drum is
flying so unfortunately I'm going to
tell you what I'm going to do with
walking conversation right now do it on
the drone and then once the Joan is done
land and it's you know landed then we'll
be able to talk together again and I'm
going to show you what Watson replied to
my phrases so let's begin so right now
what I'm going to do is I'm going to
take off the drone make it move up I'm
going to check the altitude make it go
down check the altitude once more and
then land let's try that out so let's
begin with a take off man
all right
[Music]
I dissipate go up
and that will be Atlanta bright all
right sweet now that actually went
completely and if you can see back to
the PD average said take off what we
understood my intent to take off and
said all right taking off
it was then even send a command take off
to the raspberry x who then send it to
the drill I can also tell it up sweet
and what this means is that lots we will
move the drone up at a pace of three
okay then I said stop okay
now you will begin to fuse when it's
soaked moving up it didn't land it just
went back into its own color stick how
did that happen well essentially the
stop and Lance commands are not the same
if you I mean what stop actually means
is it will actually stop quite literally
up-down and rotational movement and
notational by the way 200 clockwise or
counterclockwise all right so
essentially land would actually land the
drone we're still puts it into a new
hover state instead of allowing it to
move more all right but essentially the
difference so now it's stopping me up
it's just like hover at that new
altitude that it has reached and next
down for either down for that means of
course a human will be able to say if I
want to move the drone down at a pace
before Watson understands it as well
and so it says alright going down at a
rate of four alright then I say stop
once it reaches a sufficiently low
height and then I asked it for altitude
information and in this case it actually
got quite well so it's a 0.23 five
centimeters which was correct
all right all right your platoon three
five meters now can choose between
target for everyone that's because we're
actually in Canada right now in the u.s.
adopts we were recording by the
different system metrics yeah alright so
continuing belt that was pretty
interesting this time let's try painting
off rotating clockwise and then rotating
counterclockwise then when
shine out shall we let's take off
I think John
oh right so now the reason you saw that
stop in between and go down and
sometimes when the propellers don't want
to work it just kind of lands them sold
but it did it was able to recover and
get back in the air
so essentially let's take a look at
Watson's responses to and that landing
installing though has nothing to do with
the code correctly
okay the drums only for strong I guess
but again it's able to recover okay
that's fun let's take a look at Watts
responses now shall we so essentially
again I just take off and it says
alright taking off now when I said
rotate clockwise or if it's alright
rotating the drum clockwise any piece of
four so now one more thing I'd like to
say is that the speed in this case is
four
however it can lead from one to ten
really and it's any number between one
and ten all right so essentially one
more thing I'd like to say is this is
where Watson's powers come in so
essentially instead of me having to
manually program in every single
sentence all right rotating the drum
clock lines or the paces once all right
located in Osaka education like two
three four five interns me having to
create all of those sentences one by one
Watson conversation will understand the
entities inside of that phrase along
with your intent and after doing some
analyzation you know the pattern
recognition she's learning a lot sort of
stuff it will be able to eventually tell
me and it actually generate a sentence
all right rotating the drones clockwise
at a piece of course and it would be
able to keep repeating that like four
five six seven eight without you having
to enter in verbatim each time perfect
and that's where the machine learning
element comes into that correct yeah
Sachi and of course the machine building
element also comes in the same place
when it's deciding what you said so like
rotate clockwise for that's essential
machine learning algorithm and walk the
conversation which I'm going into in
just a little bit and that's able to say
that hey the user actually wants to
locate right now
all right so continuing them all right
now after that I stopped rotational
movement and then I told it to rotate
counterclockwise and speed of four all
right so now essentially as I said is
able to understand the entities
paralyzed and forth and the content as
rotates so it says all right rotating
around counterclockwise a piece of four
okay so again it's in to extract the
intent and entities and generate
suspension then I told to stop one more
and then I'll and withdrawal and that is
essentially a quick demo of how the
drones capabilities work but now let's
get into how the drone actually works
yeah you have like logic or so we can
see how that looks
take a look that so now if I wasn't fast
enough
what you see on my map where Watson or
input that will essentially the main
nodejs application that I have running
on my Mac all right now there's one more
critical aspect to this application the
internet now seamlessly wondering why is
it so important for nodejs to be able to
communicate with the Internet well it's
because Watson is on the Internet as a
REST API which is again what I want
Aquarius advantages you don't need a
device that has a ton of computing power
to able to use it you can use it from
any computer any device as long as
Internet exactly going to have
interaction and programming these
capabilities that allowing you to
connect to the internet
okay now the drone I mean the main
nodejs application is able to
communicate with watson conversations
via the interest okay but then we have a
problem and the problem is that we're
using a parrot ar.drone to point out why
is this a problem you may ask well Paris
decided that they did not want to use
bluetooth with this drone but instead
they wanted to use whitewashes so now
the reason that this is a problem not as
well the reason that this is a problem
for us is because well I mean first of
all when you have a Wi-Fi for interface
it can only connect to one connection at
a time however we Bluetooth as you
probably know you can connect to
multiple connections at a time your
keyboard mouse webcam microphone eating
a drone at the same time and so now we
have to somehow find a way to bridge the
drones Wi-Fi hotspot with the nodejs
main applications internet connections
and so in order to do this there's a
sort of secret sauce that I use which in
this case is a raspberry pi okay
yeah so if you didn't already know the
raspberry pi 3 has network interface
still a Wi-Fi network interface built in
now using this built-in Wi-Fi interface
I can connect to one Wiseman connection
and then I just install a new USB for
the Wi-Fi dongle occupy as well and so
now we have to in that connection okay
one is going to Jones and once the name
internet connection and so now
essentially I'm able to use the no
transmitter application to send commands
to the Internet which goes to my friend
hi which is then sent to the jones Wi-Fi
hotspot and in terms of a drone and I am
of course also able to bring information
from the drone to the Internet to the
main application where instead goes into
walkie conversations grunts experiment
which I will talk about in just a moment
now just in case you're wondering so
what it is actually works is by using an
HTTP or technically free HTTP servers on
the Raspberry Pi one listening on port
4000 one on 5001 on 5500 now if you're
wondering what these sports are well the
4000 port essentially allows me to have
a live PNG stream of the drill so
essentially it
from frieza's front-facing camera of the
drone using why screams choose export on
the 5000 port I have a JSON lots of JSON
navigation data which is being pulled
from the drone everyone seconds
essentially this data contains a lot of
stuff it contains like movement data
rotational data battery temperature
pressure
literally everything you can imagine
about the drill okay and so that's
essentially and of course that actually
goes into the contest variable water as
well and want to go to the context
variables Watson watches it quickly as I
said aware about what the drill script
battery level is in the altitudes all
little stuff of course and of course
45,500 is sort of main horse I guess
it's essentially reports that takes
ripping on the pipe for commands from
the mill jazmyne applications which
comes in turn from Watson and so
essentially this port will listen to
those commands and why is it gets firm
it'll do a little little analysis on
them and push those to the drum and then
the drug will be able to execute that
command alright so as I said I can't go
plug it into that I was just saying that
if you'd like to see the source code all
this application to actually they'll try
it out for yourself include this one
logic in your own applications please do
because right after this event this go
through the open source ultimately no
one actually let's just do that right
now why don't we so if I go to my github
page here awesome chill so it's sunny
and maybe I would like
and your your github name is tan maybe 1
2 3 yes nice so now if I go to desktop
and as well setting the wall he does
that he has a whole bunch of other
applications that he's created and these
open source that code on there right for
example like at San Mae you built on
there and also I think you've also got
some really good demos on YouTube right
ok I made that she so it's a really cool
place to where you can check out one how
they leverage his code to work and build
out your own drone or built your own
I think you did like a mousetrap IOT
device at one point yeah nice so is it
now published yes I mean that that's not
github just yet that's going to be like
it like soon though
and then I mean once it's on get help
though I will think about it so
whoever's walking right now if you want
that code it will be up to what's it are
we gonna do yeah actually I'm going to
get into how Watson conversation now I
can begin adaptive walking conversation
does a majority of pre-much all
processing in this application ok now
let's just say that your frame is go up
at a speed of 2 now a human would
actually within like a middle section
will be able to tell you that your
sentence means you want to you want the
drone to go upwards and the pace at
which do that is - okay well I'll
computer candy bastards all right a
computer does not understand natural
language at all okay however again we're
talking about Watson which is able to
introduce artificial cognition into this
and able to understand this natural
language by extracting the signal from
the noise essentially no okay this is
natural language it has a lot of noise
as a computer just stopped at me and so
we have two though pieces of key
information and these business
Meishan are up and - okay and these
pieces of key information are called
empathy and lots of conversations pain
alright so now for example up all right
when I was trimming Watson conversation
I defined an entity type and I call that
entity type up down direction now up
down direction the entity type has two
values up and down alright and in this
case it shows the value up because of
course I was a drunk to me up alright
next we have a feed every time when I
was trained Watson conversation I gave
it the speed entity type and it has ten
values 1 2 3 4 5 6 7 8 9 and 10 are its
values and in this case since I chose to
go up in a sea of two it'll say -
alright which is correct now we have the
key pieces of information extracted all
right and that pretty does alright but
we need more information to be able to
actually determine what the user means -
brain so in this case that team that key
piece of information is called an intent
all right so now what is mentioned well
technically we could just turn in what
the user means by looking at the
entities of their friends button cuz
they okay I'll go over to see - but
let's just say that you happen to say
take off and you find a face mention up
and you by mistake mention - if it were
just a certain amount of entities it
would go up in a state of - even though
haven't even taken off yet I'm so in
order to solve this problem we use
intent now just a little bit of history
about watching if you'd like to hear
this essentially before Watson
conversation is released just a few
weeks ago in fact there used to be a
service called walkie dialogue ok so
what can dialogue was essentially the
old times walking conversation alright
it helps you to create chart box but
they didn't have the intent ability it
was just the dialogue
I love flow and entities alright however
that I mean that was powerful but a lot
of people started to combine the natural
language classifier with dialogue it
would deplete very powerful chatbox and
so essentially what the guys over at
wants to decided to do and just before I
continue I'm sure you all heard of this
one of the most famous examples of this
was the newly selected sample right
where you could give it you know in a
chatbot information about the type of
movie you like to watch and then it
would be able to suggest like a movie
for you to watch okay hi remember play
nice application it was able to combine
the natural language classifier with
walk and conversation to create one of
those really powerful chat box all right
but now that was one more problem okay
so are not probably necessarily but the
gut isn't locked inside it why not
merge NLP natural a massage fire and
dialog to create conversation resent
attire dialogue alright that was a great
idea now essentially we have one service
called conversation that combines the
natural language classifier along with
dialogue that you have intense entities
and the dialogue well and of course if
you just like to use conversation like
dialogue go ahead and do it there's not
one helping you I mean you the end has
the intense and entities features or
completely optional you don't need to
use them so you can make a conversation
act like dialogue so like if then if
they respond with this just respond with
that you can make it but it's very
manual and it takes a lot of in whereas
within ten like you said you didn't have
to write out the whole sense of all the
time like if I wanted to change my
flight if I didn't say JFK if I just
said JFK and not Airport it still knows
with that accident because I'm in this
type of an app I'm looking to go to JFK
right and that's how he really does help
out so let's say
we have this rake all right now even
though we are technically sending this
race to watch some conversation to
understand the intent the algorithm
behind how watching conversations
classify it is the same as the natural
image classifier so say to assume that
we are actually sending this request to
the natural language classifier just
with a different API is that the
conversation search so these are also a
pion is right yeah these are always yes
the natural response wire conversation
dialogue needs to be one many more
powerful services that you can combine
into your application to take one very
powerful application in fact one more
thing is a little bit off topic but one
more thing I will recommend it
so the combining some services like or
even lines with conversations create
extremely powerful chat box that can
provide useful information to your users
but continuing zel so again as I said
it's safe to assume that this place is
going to the natural language classifier
except with a slightly different API
those conversations you know API so once
conversation or natural language
classifier is done classifying
your phrase you get the intent uptown so
now Watson knows that your intent is to
go either up or down
then from the equities it was able to
extract a deeper meaning and the super
ruling was instead of just wanting to go
up or down at a specific speed it knows
that speedy and it knows the direction
you want to go up and you want to go at
a speed - and that is essentially how
the entire length of this application
works for gin so we have any questions
oh yeah let me check the Facebook live
while you do that though when you show
us with Enric arrange each other's you
write a book right
yes actually huh in fact I have actually
written a book I called Helen flips iOS
app programming for kids and other these
Skinner's and this essentially funny as
the title suggests a book for kids and
beginners chorus iOS octave L and first
with from scratch so essentially I mean
if you'd like to check it out to do if
you think it could help somebody I'd be
glad to Ford to help somebody
I love to share my knowledge so this is
another one of those media through which
to do that of course I do also have a
YouTube channel where you can now see
how I like to share my knowledge and
stuff like that but also just in case
you were wondering more about Watson of
course you can actually contact me and
see my tutorial on my youtube channel
but I also recommend highly recommend
actually attending world of Watson this
year which is another one of those
concepts was hosted by IBM officially
that is about Watson time you can use
cognitive capabilities it's actually
this month in October it's going to be
Las Vegas that's gonna be very
interesting in fact a few of my mentors
will be presenting there as well and we
might have a fan in Japanese but
continuing don't one more thing I'd also
really highly recommend attending World
unimpressionist Watson developer
conference again a great conference
pretty much a cool money for developers
about Watson and this time it's going to
be in San Francisco so a couple
questions that we have actually it's
kind of funny I think we've already
answered it but again if you wanted to
use these Watson API services how would
you go do that
Watson API and that's actually one of
their high advantages is that they're
available online you really don't need
any like specific amount of computing
power in order to be able to use walk-in
services because they're actually hosted
on bluemix for you as you like services
so if you want use any of the services
that I need in my application I please
feel free to you can just go to bluemix
sign up for an account if you don't
already have one and if you do just
search up walking in real huge catalog
and you can create all these different
types of services makes them together
great really powerful cognitive
application
someone actually asked this and i don't
know we're going to solve that problem
right here on the page but I said I wish
Watson could help the IBM storage team
quickly provision storage that depends
on what was more specific problem is
technically code if you had a very
specific problem that requires coma
cognitive abilities yeah yeah I think
what you could possibly do right is if
you're trying to identify let's say
you're looking at types of workloads and
you know it's going to be you're getting
a lot of inputs on your application you
could potentially do predictive
capabilities into and recommend we maybe
auto scale the storage to be provisioned
for this so or you could even empower
Watson rights to make the decision for
you if you wanted connecting to whatever
source you're using to provision storage
is that fair enough to say can you do
that definitely I mean as long as it
applies that type of possibility yes
okay another person asks how old you are
and then we fought about it earlier but
you did I think you just turned 13 so I
would like to wish him a happy birthday
featuring 13 yesterday right yeah I know
actually also the 16 was 2 or 3 and oh
yeah I was late for the birthday that
was my fault
see what else we got over here ok mate
which one of the Watson API is your
favorite
well I wouldn't necessarily have a
favorite because they're all they're
useful in what they do they do very
different things there's a huge diverse
amount of api's and that's why it
doesn't necessarily have a favorite but
do I have to mention something in terms
of community news and it has to be a
sort of a natural language classifier
and in sort of the most sort of thing
that I could think of least case of
course it would have to be conversation
but things that I like to mash up kind
of thing will be the natural end
classifier moves even right for example
conversation or even rank are very good
couple as well and force content in
place is another very powerful service
that are actually currently creating
apps for which you will find out about
soon on my github page we have one more
question about actually the conversation
89 and the gentleman asked does it allow
for customization and what we mean by
that like initiative and converse ation
at a particular time or even setting a
context to the conversation definitely
in fact about
context cluster actually first this
project uses the contacts variable of
Watson heavily essentially every one
second I grab new data from the drone
such as battery and altitude data and I
put that into the Watson conversation
variable and so essentially in
throughout with suelto dialogue flow I'm
able to use those you know consoles
context variables in order to tell me
what to say next or he started to change
how I walk and speaks like for example
let's say you're mid-flight you want to
know that by other context variable it
could also know the amount of battery
life and let's say okay so the battery
is getting less than 20% and your
mid-flight right in that case you could
say okay Watson health user that the
battery is getting well you might want
to land now and when it gets to 10
percent to just land it without asking
the user that type of thing can be done
with the Watson context walking
conversation context variables so I know
your personal story about how you
started and you got into programming and
what inspired you and where you're
hoping to go do you mind sharing that
because I know it's the mother to watch
you're wondering you know what got you
into programming and now where are you
going happy to go yeah sure definitely
uh well physically honest actually right
I mean what got me into programming was
actually when I was five years old of
course I didn't really have that much to
do all day so watch my dad programming
almost all day and so it really
intrigued me as how like a computer
could add two numbers or display my name
on the screen and so eventually I wanted
to find out more about how it was able
to do this and how you can get a
computer to do this and so of course
after you know asking my dad to teach me
a bit of programming
introduced me to the concept so the you
know batch programming
started in batch pro yeah oh gosh well I
had actually a little bit of foxpro as
well and a little bit of C and after I
you know
fluent with the programming fundamentals
basics I started to maybe see ya there
to see the programming I started to use
the internet as a learning resources and
then I also learned how to do iOS app
development and then at the age of nine
years old my first iOS app t-table for
the acceptance into the iOS store and
t-table did your math homework correct
no actions taught you how to do until
your time sale I feel like the teachers
in your school they're probably like can
we actually use your app to do this run
so in xs/s which is another interesting
question can we integrate Watson and
JavaScript based framework yeah player
percent in fact I'm actually using
JavaScript right now in order to control
the children even ojf specifically in
son using he walk and develop a cloud
nodejs SDK which you can find on github
on the walking Elvis mile github but and
so they essentially have this like
really colorful framework to use almost
every single walking service there is
and it becomes really easy but it having
to send you know rest requests all times
you can no excuse their API and it makes
it a lot simpler although I must say the
way that I give walks and works is a
REST API is much simpler than most other
companies can provide I even more into
the sort of fun the programs when you
have a framework instead of having to
call REST API over and over again so of
course the nodejs sdk will really help
you out with that now I can speak to
this but I want to see what your
thoughts are on this first gentleman's
asking can I know how
Watson will be used to detect advanced
persistent threats because this is about
security right can we expect it to have
more human-like approach if we can
detect new threats or do we still need
upend on humans to research and write
direction-- signatures so how does lots
and fall into play around security and a
logs and central attacks Brett well um
it depends on your thinking about sort
of technical security like falsification
yes
not with physical security so well as
far as I know I don't think I'm a flying
saucer might be going here and not sure
if it will be powerful enough for this
type of a new space because I mean as
you can see this sector could be very
diverse there will be lots of different
issues rocks caused by wattage of the
problems relationships between men will
be very hard to determine unless you had
an absolutely gianormous training set
that in which wanting to find tires from
so I'm not exactly sure how you would go
about doing this with walk in that wave
but if you could implement some order
custom I mean with such advanced
problems you don't need a special
walking service I mean there are other
walking services that aren't available
on the public bluemix package but you
can pay per month to you and so those
more powerful walking services may be of
more use for the type of scenario or
maybe even a costume neural network
might work yeah and I am what I could
say is actually that IBM security is
actually working with wants and
capabilities and coming out with
potential offering is not already on
identifying the patterns and trends and
intent right because that's one of the
most powerful things about Lawson right
is to identify patterns and trends and
the context are you're relating to that
right and if you start training this
type of a influx in your network it's
probably coming from this type of
situation and therefore we should
protect ourselves in that or is how we
respond right someone also asked how
accessible is Watson for someone who
does not have a technical or programming
background in fact actually just before
I continue that is again one of the huge
advantages of IBM Watson you don't need
a like a PhD in machine learning to be
able to implement the sort of stuff
that's the best part all you need to do
is have some fun you know small
programming experience like I mean a
success you I mean I heard your question
you don't have that much a programming
experience but if you can get started
you know learn programming language
which in case you want to undo so the
iOS development course I said my both
tell us this could help you out with
that and so essentially you just need to
learn that sort of basic coke
fundamentals and once you are able to
get to that to the point where you can
create applications there after that
it's literally just a few rest API calls
to call Watson and to be able to create
your own cognitive based application so
you're going to know that whenever here
the thing is interesting and that was a
great explanation of how we can use it
actually again I'd recommend going to
IBM Watson or bluemix.net and looking at
which of the services interest you
because we actually have very nice user
interface examples of understanding how
they would work and you could
potentially leverage those in your
day-to-day or your environment and then
what you want to get more into the
programming aspect we even had SDK kits
for each of these services available to
you on to help you controlled us another
one person said will Watson be cost
effective to reach low budget clients so
you know I people like Walmart could
probably afford it but what about the
API is is going to be you think that
like if I want to start my own company
could I leverage it or I really do think
so and in fact again that's one of the
advantages about Watson it's not
overpriced Watson ears relatively
affordable for small businesses yeah and
so I'm going to tag on to that one too
I've actually worked with a couple
startups in New York City I'm actually
one of them with boom fantasy and they
are using Watson trade-off analytics to
identify with their market budget every
month whether or not they're receiving
which of the customers or consumers or
the ones that they should be marketing
to because as a start-up you know your
budget is limited right and you have a
you have that burn rate that you have to
deal with so if you can validate every
month or even sooner how efficient your
money is being spent on the marketing
notice through YouTube whether it's
through you know you know meetups or any
other types of campaigns that you'd run
that'll help you out so it's definitely
used in the startup environment all the
way on to the large corporate companies
that are using it today and again one of
the things that we work with -
especially if you're a small start-up
looking for that about IBM Global
entrepreneur program we'll be able to
provide you credit to be validate for
that so you'll be able to
leverage our services on bluemix and
watson another thing that i was going to
go act here someone actually said kid
watson speak captcha image so you know
the caches online screening would it be
able to know to go into and say we were
River it right no I mean in theory yes
but then again since some capsules are
coming Li Randall it may take a lot of
training but then again it's completely
possible using visual recognition and of
course the text-to-speech then again a
lot of money would be acquired and I
know a lot of folks are actually
continue to keep asking about the price
and how much it's going to cost again
the way we built these initial AP is out
for the public to test with you can sign
up for your bluemix it's a free trial
then from there what you can do is
you'll see the API calls capabilities
and actually if you're going to ask
you've hit the premium model if you
actually want to start going past that
again there's always multiple ways
potentially increase your credit if you
validate for that through IBM global
entrepreneur is Nespresso you can work
with you on that as well another thing
that we do is you do see regular credit
card and you pay per API call so that's
actually again how it's working that's
where we see actually cloud going you
pay for what you use exactly right
and what your and we're going to provide
a link to Panama is a github again I
know you're wondering though its kind
maybe one two three I cannot pull that
up right now so this is my github page
and you can see all my open-source
projects here now I want one more but
Tammy what do you uh see yourself going
next after this well um to be completely
honest I think it would be well mainly I
want to continue to do what I'm
currently doing alright so to continue
you know my fossil development keynote
speaking iOS app development opens it I
mean of course continuing to share my
knowledge and of course those with
cognitive like Watson
I created cops and machine learning
solutions as well I want to continue
using things like a I for example and so
you know I want to continue to gain more
experience at this and as I gain more
experience or share my knowledge about
it and you know just keep getting better
reaching out to more people in fact
actually one of my goals is to reach out
to and help at least 100,000 aspiring
kid coders and beginners so you speak at
stem events right as well you've done
that before and I think you're going to
be in Northwest Bentonville Arkansas
speak to a couple of students here I've
spoken to schools in Toronto as well so
I know the again I know the answer to
this because I'm forced to be it was the
March part of the marketing a prefilled
aspect but people are wondering you know
how can I be M assist self-driving cars
automobiles autopilot self-flying
vehicles boats in Union anything that
has a motor on it how would they help
well that's actually what my toll was
the crazily
like an autonomous drone that utilizing
Watson and I really think IBM Watson
provides very colorful capabilities for
you to be able to do this type of thing
so for example actually visual
recognition can really help you out with
saying these lights on like let's say
you're building an autonomous RC car you
can use visual recognition to see
whether the drone I mean with the RC car
keeps light and start turning slightly
little or slightly right slightly left
as less how visual recognition can be
used in things like RC cars drone for
example it can detect faces all that
sort of stuff I can really reuse in
autonomous vehicles very nice so an
example that we're actually using this
today is a thing called Ollie so Watson
ollie one it'll be at world of Watson
but we're actually implemented it
there's a 3d printing company that they
printed out all the parts in the span of
icing was like three months or less
massives creative and they built with
Watson as the brain behind it speech the
tags text-to-speech so with a person
person the button says hello how can I
do or where should I take you and you
could say I want to go to the Italian
restaurant where is good and they could
recommend list and then I'll start
driving for you along
that has it's actually in National
Harbor in new on Maryland there the
Washington DC area and they're using it
right there I know we're actually
looking at leveraging and other
implementation as well so that's
actually where and how we've already
been using this in self-driving
automobiles People's Republic
transportation let's see what else we
got here let's see what G okay great I'm
curious what what you want to do in life
did you already say that yeah on the
outside and if you want I can repeat
that sort of I mean as I said my goal is
to mainly reach out to and help at least
100,000 silent encoders and again it
works actually I'm already around 2,000
people there and the force uh continuing
with our reach cycles and higher one and
continue to share my knowledge and help
as many people as I can and that's
another thing please like and follow him
on Twitter and Facebook and YouTube it's
not already I mean because he's not
stopping here guys he's going to keep
doing more is working with all those
audience top minds like John Cohen and
markets around I owe t another quick
question though and I get this a lot too
is is IBM Watson and all these AP is
another agrestic rest like guys could I
could I use it with an application
that's on another platform that's not
the bluemix so which is like AWS or
Microsoft or as far as I know Watson
does not provide that capability yes as
far as I know yes but for now though you
have to host a job for the Watson
services on do my services off limits
but then again
so ease of use of Lumix really just
really tells me that there really isn't
enough to need for hosting I've been
walking services on other sort on the
platform okay so one performance always
feel free to do try it out on bloomix
you know that's how we're going to host
this our platform as a service and there
you can combine it with a lot of other
things if you're an enterprise you can
start combining it with your business
process managers your systems of record
because we're going to expose it through
API connect and these are now going to
be restful api you can now combine with
watson valuable insights that may be
your amusing mainframe on the back end
these apps that never been able to touch
it a new capability that's right that
you can apply that to have some pretty
awesome stuff let's see how did you
describe the services between oh yeah
someone mentioned how would you describe
the differences between isis amazon echo
and watson okay have you played it all
the other one i go at all that was when
i go yes so this is just my preference
or point of view when it comes to it IBM
cognitive capabilities we've really
designed them to be enterprise grading
scale we've tested them mattias lee
we've built these corpuses up at a
certain level so when we give them back
to you as the user they're going to hit
a certain quality assurance level right
off so one of the things that we don't
do versus some of the other like Google
now or Siri they ask you and for your
data to them be shared with others
because it's part of the training
process we don't do that because we want
your data to always be your data and so
that keeps your competitive advantage
right if we were to have all the banks
give us the data to train and give it
out as a platform
you've lost your citizens so on that
note Sam a is there anything else you'd
like to share before we leave I think
that we've already talked about what a
lot an interconnect and I guess that's
actually going to be it for this
Facebook live event again thank you
everybody for tuning in and yeah I mean
if you'd like to check out the source
code of this project please do it's on
my github can maybe want to sleep there
will be a link to it after this event
and then yes looks good I mean if you
have any questions you can actually
email me am I
mail as many gmail.com that will be in
the description with Facebook Osment as
well and I think that's gonna be it for
city alright thank you again for tuning
in and again go to his website go to his
Facebook go to github and to be doing a
lot more with this on and on and good
something thank you guys goodbye