Please Subscribe Like Share Comments
SAGE TALENT QUEST 2018
Đăng ký:
Đăng Nhận xét (Atom)
-------------------------------------------
Alec Baldwin Gave Melania A Nasty New Name And Trump Isn't Happy About It. - Duration: 5:14.
Alec Baldwin Gave Melania A Nasty New Name And Trump Isn't Happy About It.
Alec Baldwin apparently had someone read to him an article that was published by the Internation
Business Times which claimed that Trump's Consumer Financial Protection Bureau chief
Mick Mulvaney tried to get the bureau to "back off" an industry that had "bankrolled
his congressional bids."
He then attacked the Trump administration by tweeting, "It's pirates, whores, and
thieves holding every type every of resumé in this administration.
But it can't go on forever."
The funny thing about his tweet is that as reported by Breitbart News Baldwin fails to
mention that Mulvaney made his requests to the CFPD when he was a Congressman and way
before President Trump ever appointed him to head anything.
This attack comes on the same day that rap artists Kanye West and Chance The Rapper set
social media on fire by sticking up for President Trump in an effort to defend the rights of
those who have different opinions than the liberal thought police which has become a
big issue on social media.
This is not his first verbal attack.When Baldwin's daughter, with Kim Basinger, was 11-years-old
he called her a "Rude, thoughtless pig" in a voicemail he left to her.Baldwin seems
to show a pattern of disrespect as seen here in this most recent episode via Fox News:
"Alec Baldwin has apologized after he used homophobic slurs to violently threaten a British
reporter who said his wife Hilaria Thomas had tweeted about wedding presents and TV
appearances during James Gandolfini's funeral on Thursday.
"If put my foot up your f**king ass, George Stark, but I'm sure you'd dig it too much,"
one tweet read.
"I'm gonna find you George Stark, you toxic little queen, and I'm gonna f**k you…
up," he added.
Baldwin denied Thomas had tweeted during the funeral, and said the tweets were instead
made after the New York City service: "My wife DID NOT use her phone, in any capacity,
at our friends funeral.
Now, f**k this twitter + good luck to all of you who know the truth."
The 55-year-old actor's Twitter account was disabled following the rant.
"When someone tweets something at a certain time, and you retweet it later, it posts the
TIME the first person tweeted," she explained on her own Twitter account later on Thursday.
A rep for Baldwin also stated that Thomas did not tweet during the service.
"It's disgraceful that a reporter would manufacture and publish a story and not call
for comment or explanation — especially when it's intended to take away attention
from honoring the memory of a beloved figure like James Gandolfini," the rep stated.
"There are multiple witnesses to the fact that Hilaria left her phone in the car and
wouldn't, couldn't, and didn't tweet during the service."
When asked specifically about Baldwin's homophobic tweets, his rep resent FOX411 the
above statement, which did not address the nature of Baldwin's rant.
Later however Baldwin wrote the following to GLAAD:
"My ill-advised attack on George Stark…had absolutely nothing to do with issues of anyone's
sexual orientation.
My anger was directed at Mr. Stark for blatantly lying and disseminating libelous information
about my wife and her conduct at our friend's funeral service.
As someone who fights against homophobia, I apologize."
Baldwin appears in a series of TV ads for Capital One bank, promoting their credit cards.
A rep from Capital One did not immediately respond to FOX411's inquires on whether
or not Baldwin's latest rant may affect his role with the company.
Dan Gainor, vice president of business and culture for Media Research Center, told FOX411
he doubts Baldwin's endorsement deals would be impacted by his tweets.
"This is the kind of incident that shows the hypocrisy of the liberal media.
Paula Deen says something bad decades ago and has her whole career destroyed," he
said.
"Alec Baldwin fills his career with embarrassing moments – saying awful things to his daughter,
getting kicked off an airplane and now, the ultimate PC sin, anti-gay comments.
And his career will continue unharmed."
Indeed this is not Baldwin's first tech-based tirade.
He was thrown off an airplane in 2011 for refusing to turn his phone off while playing
Words With Friends.
And following a Feburary altercation with a newspaper photographer, Baldwin reportedly
made racist remarks, calling him a "coon" and a "crackhead."
"He was saying some serious racist stuff," New York Post photographer G.N. Miller said
at the time.
"He said some choice words about my mother, and he was telling people in the street that
I'm a drug dealer."
Baldwin denied he made those statements."
There seems to be a pattern with Baldwin's verbal comments.
We will just have to wait and see if he apologizes this time.
what do you think about this?
Please Share this news and Scroll down to comment below and don't forget to subscribe
USA facts today.
-------------------------------------------
How and when do you talk about how your relationship might change as MS progresses? - Duration: 0:53.
For more infomation >> How and when do you talk about how your relationship might change as MS progresses? - Duration: 0:53. -------------------------------------------
Should investors be cautious about ETFs? - Duration: 5:56.
For more infomation >> Should investors be cautious about ETFs? - Duration: 5:56. -------------------------------------------
Soy Luna 3 | Yam wants to forget about Ramiro (ep.9) (Eng. subs) - Duration: 1:29.
You're here.
Where were you?
After Blake, I saw you running and I didn't see you after.
Are you still mad?
No, I was doing something with my mom, she asked me to help.
Are you sure?
Because you didn't talk to me all day.
That was because I was a bit mad.
A little?
You ignore me.
I think that it wasn't that bad what I did.
I only accepted a mini advice from Ramiro.
Don't say that name anymore.
What do you think if we rehearse?
Why would we loose an incredible opportunity because of Ramiro.
The guy that left us for the Red Sharks.
Past.
We have to put us into our vision.
Perfect but on one condition.
We'll never give up. Ramiro is part of my past.
But for the last time.
Don't you think you're a bit too hard on him?
I think that for your ex.
Skating is very important.
It's the vision for us.
I think you're exaggerating a bit and maybe you need to put yourself into his shoes.
I have nothing to understand.
He made a decision and he was wrong. There's no way back.
-------------------------------------------
Teachers, Students Enraged About Growing Rat Problem At Newport Harbor High - Duration: 2:03.
For more infomation >> Teachers, Students Enraged About Growing Rat Problem At Newport Harbor High - Duration: 2:03. -------------------------------------------
What you need to know about MSI B360M MORTAR | Gaming Motherboard | MSI - Duration: 1:42.
This is the B360M MORTAR motherboard.
MSI's newest affordable Gaming standard.
It features a complete new design with 26 percent increase of the heatsinks
To dissipate the heat even better for the new Core i7 and i5, 8th Generation 6-core processors from Intel.
So all the MSI B360 motherboards also support PWM and DC mode Fan headers.
Which means you can connect any type of fan to this motherboard and control the fan speed.
It also features unique Gaming Features such as Audio Boost.
To give you a more immersive audio experience when you are playing your game.
Fully isolated from the rest of the motherboard with a 7.1 output.
Besides audio design it also has MSI Gaming LAN.
Optimized to giving you the lowest latency when playing online.
And for storage, the B360M MORTAR has 2 onboard M.2 slots
for fast NVMe based SSDs for blazing fast loading of applications and games.
So it means the B360M MORTAR supports the latest M.2 devices on the market.
It has 4 DIMM slots using DDR4 Boost Technology
This makes sure your RAM is always supported, no matter the speed or the latency.
And for video outputs, it even has onboard HDMI and DVI (+ DisplayPort)
And to complete this motherboard it also has plenty of SATA connectors
USB 3.1 Type C and Type A
Making the B360M MORTAR the perfect affordable Gaming solution.
-------------------------------------------
History About Our Channel | 1 Year Celebration | Poda Dubukku - Duration: 2:36.
For more infomation >> History About Our Channel | 1 Year Celebration | Poda Dubukku - Duration: 2:36. -------------------------------------------
Trey Gowdy EXPOSES The Truth About Comey It's Not Just A LEAK It's A FELONY(VIDEO)!!! - Duration: 6:10.
Trey Gowdy EXPOSES The Truth About Comey It's Not Just A LEAK It's A FELONY
good evening at welcome to tucker carlson tonight rapper Kanye West has
spent four days posting pro-trump content on Twitter what does that mean
for America and humanity and life itself we'll investigate just ahead but first
Fox's Bret Baier just conducted a deeply revealing interview with former FBI
director Jim Comey during that interview Comey tried to claim that the steel
dossier the one he hid information about from the president was not central to
the request for a FISA warrant against people affiliated with the Trump
campaign here's part of it called the dossier unverified salacious why did you
use that to define a court to ask for surveillance for Carter page not only
use it but you led with it a bulk of that FISA application deals with that
dossier why yeah that's not my recollection Bret and I don't know that
the FISA application has been released my recollection was it was part of a
broader mosaic of facts that were laid before the FISA judge to obtain a FISA
warrant it was a lot more than the dossier in the FISA application my
recollection was there was a significant amount of additional material about
Paige and why there was probable cause to believe he was an agent of a foreign
power and the dossier was part of that but was not all of it or a critical part
of it is my recollection I don't know that it's been really hiding behind
government secrecy once again congressman trey Gowdy represents South
Carolina he chairs the Oversight Committee in the House of
Representatives and joins us tonight mr. chairman thanks for coming on what did
you you want to know if you what did you learn from it a couple of things number
one the page in straw he would have fired immediately which means the two
ages that were leading both the Clinton and Trump
investigations should have been canned director Comey's recollection is flawed
if he does not remember telling Congress that his agents told him they didn't
think flem was lying then he needs to get his lawyers to go back and look at
the transcript we did not miss here maybe he misspoke but that's in the
transcript and the double standard that he's had for the last couple of weeks
still exists you know Clinton can lie and she ought
to be President mccabe can lie and he's still a stand
god the Trump boy if he tells any lies at all then impeachment is too good of a
remedy for him and that dangerous Martha Stewart who went to prison for lying and
he brags about that in his book right he doesn't remember a lot he doesn't
remember when he first heard about the Steele dossier
he doesn't member who told him about the Steele dossier is that plausible no
because he knew that Republicans paid for it even though that's inaccurate so
if he learned about it in the fall of 2016 then it was already known that the
Democrats had picked up on that work so whoever briefed him why would you just
say the Republicans started it but but not also include the Democrats finished
it so that that and and the the part about it not being part of the FISA
application not a significant part then Tucker why did you use it at all and
I've actually read the application and he's just wrong it was an indispensable
part of the application period so how can he claim otherwise I mean his first
reaction was I don't think that's been made public in other words you can't
prove me wrong because you and the rest of civilians asking these questions the
American citizens here don't have access to the relevant information but as you
just pointed out you do you've seen it that's not true this forms a central
part of the requests correct well there are three parts of it this is one of the
parts they lead with it and if my recollection is correct it occupied a
disproportionately large part of that application so I go back and ask all
those who say it wasn't important then why did you use it if you didn't need it
why did you use it and why did you lead with it so let's get let's strip away
the politics and take Trump out of this so we can see clearly at the core is the
allegation that Carter page the hapless Carter page was acting as the former
director said tonight as an agent of a foreign power is that a plausible claim
do you think any reasonable person looking at the available evidence could
reach a conclusion that Carter page is a Russian agent it'd be tough I think most
people that have any experience with Carter page which would include the
because they've interviewed him on a number of occasions he's would you'd be
more likely to call him a clown than a spy so what do you make and I don't say
that I don't say that to be disparaging towards him I mean I spend a lot of time
with him in a congressional committee room if anyone needed his Fifth
Amendment right to counsel if anyone needed a lawyer while he was talking to
Congress it would have been coroner page I just don't I mean he's not James Bond
he's not Jason Bourne he's it's sad actually
so if our top law enforcement official can say with a straight face on live
television that Carter Paige is a foreign spy and represents a threat to
American national security he's lying it sounds like well I don't quarrel with
the FBI wanted to run out of lead I mean I I'm probably biased I am biased
towards law enforcement go ahead and run out the lead
the challenge then becomes when you are going to spy on an American citizen it
needs to be more than a lead you need to have probable cause and make all proper
representations to the court the other thing I will tell you Tucker that I
learned tonight is Jim Comey has a definition of the word leak that no one
else in house what he says is a leak is what the rest of us call a felony
leaking is disclosing a confidential conversation which is exactly what he
did he came off as much oil ear in this interview though thank you god bless you
and God bless america
-------------------------------------------
Colorado politicians speak about education rally - Duration: 0:53.
For more infomation >> Colorado politicians speak about education rally - Duration: 0:53. -------------------------------------------
Korea Expert 'Cautiously Optimistic' About Peace Summit - Duration: 2:10.
For more infomation >> Korea Expert 'Cautiously Optimistic' About Peace Summit - Duration: 2:10. -------------------------------------------
Fashion & Free Trade: Questions about Globalism, Tariffs, and Trade - Duration: 8:39.
And the world getting smaller, I think, fashion encapsulates it more than any other industry.
Cotton grown in Arizona might be spun into yarn in Pakistan, and then made into fabric
in China, and sewn into apparel in Vietnam, and sold back at a retail store in Arizona.
People don't realize that fashion is truly a global industry.
All around the world, there is fashion being created, and even if it is created in one
of the places that we're more familiar with and we think of as global fashion capitals,
it is likely to be manufactured in many, many other places.
The real-life impact here in the United States is that about 20 years ago, 25 years ago,
we had about one to 1.2 million apparel workers, the vast majority of those jobs went offshore.
I think by far, it was the cost of labor and being able to produce in a cheaper country
that made the textile jobs leave the U.S. for the most part.
So, what happens is labor costs go up, so prices go up.
The industry goes and looks for the next place to develop the industry, where it can get
products at a lower price.
Ninety-eight percent of American fashion, high-end and low-end, is manufactured overseas.
Globalization has resulted in the virtual disappearance of garment manufacturing in
New York City and in most places in the United States.
U.S. workers are frustrated.
They have been the sacrificial lamb on this altar of globalism and unfettered free trade.
Import tariffs or duties fulfill two objectives, at least in theory.
First, is to protect American jobs from imports, American manufacturers.
And the second is to raise money for the U.S. Treasury.
Today, it's not so much the main focus of revenue, 'cause our income tax and other taxes far exceed what's
raised in tariffs, but it still remains a policy to try to protect domestic industry.
A tariff is a tax imposed at the port, at the border when the truck or the ship enters
the United States.
The interest of the consumer is important.
I'm a consumer, I like getting a bargain, but you have to balance with your responsibility
to your workforce and to the betterment of human society overall.
If another country can produce apparel more efficiently than the United States and we
can do other things more efficiently here, it makes sense to allow the American people
to buy clothing at an affordable price and free up their own cash to spend on other things.
And for them to do more jobs that are more suitable to the United States.
And if we're looking to protect an industry or we're looking to help people with retraining
to keep their jobs, then I think we should have retraining programs, rather than the
indirect of saying a high tariff is gonna help someone keep their job in a, a manufacturing plant.
We think that there ought to be reasonable constraints which incentivize U.S. retailers
and brands not to buy product from manufacturers in certain parts of the world who abuse their
labor force, who abuse the environment.
Someone around the world is, generally, making a product for cheaper than what it should
cost to retail to you,
and, therefore, is cutting some corner in order to get to that point, until consumers recognize
that there is a hidden cost to what you buy.
Fashion is made possible by trade.
There are many countries that are using fashion in ways that promote economic development
in some of the least developed countries in the world.
And that's really exciting.
A rational trade policy that says, absolutely, have access to our market, inject competitive
forces and discipline into the system, but not through unacceptable practices.
Not through labor exploitation or government subsidies.
We've not lived under any significant restrictions for quite a while now.
The industry, the globalization has been complete.
The United States has free trade agreements with 15 countries in the world.
The rules for apparel on each free trade agreement are, are different.
The rules are just extremely complicated.
Free trade agreements with individual countries sounds great, but to be honest with you, it's
not as great a deal when we look at it from a company perspective; because when we do
the separate deals, everyone is different, but once you have those differences, it becomes
harder for companies to then do business in multiple free trade agreement areas.
Everybody makes fundamental arguments that, wow, free trade agreements are good.
Everybody wins.
We open our market, they open theirs.
Well, if your market is the size of the state of Kentucky, and that's what you're giving
us, and in return, you get access to 320 million American consumers, who have a gross domestic
product of about $18 trillion a year, that's not a fair deal.
I think virtually everyone would probably agree that trade rules should be simplified.
That it would be good for the economy, good for the fashion sector.
The fashion industry was impacted big-time when President Trump pulled out of the Trans-Pacific
Partnership.
The Trans-Pacific Partnership really offered an opportunity for Vietnam and Malaysia and
other countries to become a part of a free trade agreement with the U.S.; and therefore,
be duty-free to the U.S. market.
We would say, if you truly want to help this hemisphere, you have to be very cautious about
what you do in Asia.
Especially in our sector, because once you give a player in Asia preferences equivalent
to NAFTA or CAFTA, there's going to be damage for the NAFTA and CAFTA countries.
They are competing directly with those counties.
So, NAFTA, the North American Free Trade Agreement, that the free trade agreement that went into
effect in 1994, and it's between the U.S., Canada, and Mexico.
CAFTA is the Central America Free Trade Agreement.
It's the Central American countries and the Dominican Republic that negotiated a free
trade agreement with the U.S.
The Central American countries, collectively, might be the second largest source of apparel,
uh, to the United States and the world.
NAFTA changed everything.
Everything started moving to, to Mexico, 'cause it was cheaper labor.
So, we started doing fabric down in Mexico.
China opened up an entire new world, and it all became price-driven, because no one else
could compete.
We still allow people to buy product in Asia under some fairly abhorrent circumstances,
and that displaces any benefit that would naturally have accrued to Mexico or Central America.
We see the Tazreen factory fire and the collapse of Rana Plaza in Bangladesh that has generated
lots and lots of attention.
The retailer wants their orders and their merchandise when they want it.
They don't care what it takes for me or my customers to get it there, nor should they.
There are going to be, at the super low-cost level, where people are going to the cheapest
factory no matter what and don't have a good system of human rights compliance audits;
then, that risk is certainly gonna take place.
What I think most consumers don't really realize is how global our industry is and how important
it is that we're global.
The trade agreement regime, systems like quota, tariffs, that has more potential than anything
else to impact the industry.
If we're gonna negotiate agreements, let's make them meaningful and not make them so
restrictive that they're not actually useful.
The American consumer has an advantage today that it didn't have 20 years ago, and that's
the vibrancy of the internet and the vibrancy of information and transparency.
Invest some time into understanding the brands.
You need to know whom to trust by doing your homework with respect to looking at the actual
product or knowing your retailer.
Forget what they want you to believe.
Reward the people who are doing the reasonable and acceptable practices, and punish those
who are not.
And there is no greater punishment in the retail sector than to not buy something.
-------------------------------------------
Meet The Instructors - A Chat About AI - Duration: 19:11.
>> Hi, Ivan. Hi, Daire.
Thank you so much for joining me
here in this gorgeous space that
is one of our tree houses here on Microsoft campus.
And I thought I'd bring you together here so we can have
a casual chat about the space, A.I.,
Computer Vision, your background and let the students
and viewers learn a lot more about what we're doing.
So, with that maybe Ivan
you'd like to tell us more about yourself.
>> Sure, Andrew. It's great to be here.
My name is Ivan Griffin.
I've been a Professional Software Developer
for 20 plus years.
My background is in Computer Engineering and then I went
back and did a Ph.D. in Computer Science.
I'm really enthusiastic about
Artificial Intelligence and
particularly Computer Vision applications.
>> Great to have you here.
>> Thank you.
>> Daire, how about you?
>> Hi, Andrew. It's great to be here.
And like Andrew just said
my name is Daire, Daire McNamara.
I've been a Software Developer for about 20 years.
I'm working on the all sorts of projects.
I did a chemistry degree initially,
and then a Masters degree in telecommunications.
And about five years ago,
with Ivan here, I founded Emdalo Technologies.
>> It's great to have you here guys.
>> Thank you.
>> So Daire, why don't you tell our viewers
today a little bit more about what Emdalo does.
>> So about five years ago,
as I said, Ivan and I founded Emdalo Technologies.
>> Okay.
>> And what Emdalo Technologies is all
about is Artificial Intelligence,
Machine Learning at the Edge.
>> Okay.
>> And we saw a lot of people are doing
a lot of work in the Cloud and on GPU's.
>> Okay.
>> Driving fantastic algorithms,
but where we thought we could add value is,
if we could take those algorithms
or derivations of those algorithms and get them
working on smaller devices on the Edge of the network
maybe where you cannot have
connectivity to the network for instance.
>> Sounds like really exciting stuff and tons
of excellent projects I'm sure you're working on.
and busy, kept busy with
day to day. How about you Ivan?
>> Yes. So as Daire mentioned,
Andrew, we build
intelligence systems solutions for our customers.
>> Yeah.
>>The types of things include: Gesture Recognition,
Visual Asset Inspection, Room Occupancy
Detection and most of these involve
Machine Learning and, predominantly, Computer Vision.
>> So Ivan, yourself and Daire
have mentioned Machine Learning at the Edge,
and I've seen on your website,
just so we're very clear for
the viewers we're not talking
about that wonderful guitarist from Ireland,
and we're not talking about the Microsoft Edge browser.
>> No, that's right Andrew, we're talking about
AI applications closest to where they'll be used.
So, in situ in the local environments.
>> Got it, got it.
>> And that brings into
consideration for special characteristics.
One is the compute.
So typically, the AI solution at the Edge will
have less compute than if it
was implemented solely in the core network.
>> Got it.
>> You might have power considerations.
So these devices can often be battery powered
and be designed to last
for a period of time and battery alone.
You also have, then, the network bandwidth issue.
So quite often you want to take
all the data coming from your sensors and
distill that into denser knowledge
before it reaches and overloads the Core Network.
>> Great.
>> And the final consideration is latency.
>> Yeah.
>> If your in an autonomous vehicle,
you don't want to wait for
the message to go back to the core network,
have it make the decision and come
all the way back to you to decide you need to break.
>> I want to break immediately.
>> Absolutely.
>> Got it, okay.
>> That's fascinating.
So, so many applications will live
on that Edge and that's what you guys are involved with.
So our course, Dev290x is
about Computer Vision and Image Analysis.
Daire, if I was to ask you,
from a 10,000 foot view,
an overview, what does Computer Vision really mean?
>> Well, I think Andrew,
when I think about Computer Vision
I have to actually back
up a little bit and I
start with thinking about human vision.
>> Great.
>> Because you can see the human eye is
such an incredibly complex thing that it is
capable of doing so many things
that we take for granted day to day.
So that when you come into computers inevitably,
invariably you're often very disappointed by why
computers typically can't do what the human eye does.
>> Interesting.
>> So in computer vision,
what we look at is the type of
artifacts that computers see,
which is lines and lines of
pixels and all these kind of things.
And then we discuss what is possible
for computers to see, as it were.
And that's what I think of as Computer Vision.
>> Great. I think in the course will show,
it's come a long way as well.
>> Absolutely.
>> Ivan what do you think? What, if you were to -
Yeah, I think in the course,
we introduce you gradually to
the concepts of some basic image analysis where we
look for extract certain features
from the image such as edges and corners,
and then we show how you can
bungle these up and implement
very simple object detectors when you're
looking for specific things in an image.
And ultimately, we get to kind of
the current state of the art,
which is using deep neural networks to really
understand in detail all of
the elements in a computer image.
>> Fantastic. Computer Vision in a nutshell.
So Daire, we've put a lot of design into the course.
I hope that comes across for our students.
And so I just want to spend a couple of minutes
walking through the core structure and what we've,
the fantastic program we've got set up for them.
>> Yeah, Andrew. Like you say,
we've put a lot of thought
into how we've designed this course.
We see it as having three kind of broad areas within it.
In the first one, what we look at is very basic features.
Things like Edges and lines of images.
>> Of images.
>> Yeah.
>> Exactly. Just very, very basic features.
And we show you some of
the very powerful techniques that you can
use right there and then on those.
And then, we lead into a part of
the module that we call the shallow learners,
and in here we build on
all these little features that
we use in the previous module.
>> Yeah.
>> And we start using that.
What used to be thought of as Machine Learning,
but nowadays, it's kind of seen as maybe a little bit.
>> Classical.
>> a little bit classical nowadays.
>> Yeah.
>> And that's kind of where you'd see something like
the Histogram of Oriented Gradients
where you can detect pedestrians
and things like that and scenes.
>> Right.
>> And then finally, we start moving
into the more contemporary algorithms out there.
The Neural Network based approaches and you can
just see the power of these algorithms.
They are just extraordinary.
>> Great.
>> We will show you how to write
code that pulls off these tricks.
>> And we've done I think, it's fair to say,
we put a lot of effort into
making this a practical course, Ivan?
>> Yeah, absolutely Andrew. So we
find one of the best ways to
get stuck into this is not to be
too afraid of the theory,
but actually to get your hands dirty with labs.
>> Yes.
>> So get straight in there, get coding.
>> Get lots of quizzes, lots of labs.
>> Yes.
>> Ivan, in the course,
we use some different libraries,
frameworks, and a couple of toolkits.
Maybe just talk to why we went
that direction and what they are?
>> Yes, Andrew. So, the labs are
primarily written in Python,
and we use some libraries
basically to follow the mantra of
not reinventing the wheel.
>> Great.
>> We want people to become practitioners,
to get stuck in, and get their hands dirty.
>> Yes.
>> But to leverage lots of
the work that was done by clever people and spent
many months and years optimizing
these algorithms to run on hardware effectively.
For example, we use "openCV",
which is an image processing
and basic object detection library,
and we also use
the Microsoft Cognitive Toolkit
for our deep network stuff.
>> Great.
>> And the Cognitive Toolkit, in particular,
does a great job of scaling from a single CPU to
utilizing systems with multiple GPU's
to scaling across multiple, multiple systems.
You just don't do that again from scratch.
Leverage this hard work.
>> Right.
>> And learn the basic concepts
involved in the Computer Vision applications.
>>Great. And I see a lot of parallels
from the software development field.
If I was writing an app,
I wouldn't be asked to build a compiler first.
We take advantage of those who've come before us,
we stand on the shoulders of these giants and that's
why we use these toolkits and these frameworks.
So, Ivan, in the course,
we made a decision early on that we're going
to work with the Microsoft Cognitive Toolkit,
and then, particularly, for
the later modules: Module 3 and Module 4.
How did you find the tool kit
and what was your experience like?
>> We had a great time using it Andrew.
It's a relatively recent arrival
on the free toolkit scene for Deep Learning,
and it's managed to learn from
some of the mistakes that other toolkits might have made.
The API in particular is high-level enough that you
can get started and
understand the concepts pretty quickly,
but low-level enough that it
gives you a whole pile of power.
So we had a great time using it.
>> Fantastic. Daire, anything to add?
How did you find it?
>> Andrew, we found it great.
And you know I just wanted to say that we
liked it so much that we've kicked off a project,
an actual real life project.
>> Oh, great.
Fantastic. Well, keep us posted.
I'd love to know how that goes with the toolkit,
and I'm glad you found it very useful.
So, Daire, change is happening
rapidly in the space
AI in terms of everything that is going on.
But in terms of breakthroughs,
what do you consider to be the next big breakthrough
in this space.
>> In a word, Andrew, I would say, "democratization".
>> Really?
>> Yeah. I think the last few
years has all been about a group of
experts building better and better algorithms
and I think the next few years is going to
be about a lot more people
getting involved in AI and applying
these algorithms into their own domains
and into their own problem spaces.
>> Interesting. So, democratization in terms
of who gets to work in the space.
>> Yeah.
>> I kind of agree.
Just to add to that, I think what we're saying here is,
the reason we do these kind of courses is
21st century job skills,
and thinking about what people will
need from a skills bases.
AI shows up and a lot of other topics.
And to get more people into this space,
we all can't live in that top level of
expert practitioner-wise we're bringing more people in.
So, great way to put it.
Democratization of AI.
Fantastic. Ivan?
>> Yeah. So, I agree that we need
a lot more practitioners to actually take those skills
and deploy them to the real world
and solve real world problems with applications.
I also think from the technology slant,
something like Geoff Hinton's capsules idea.
Geoff Hinton is a bit of a legend in AI circles
and his capsules idea
is a new idea that's been a long time coming.
He's been working on it for nearly a decade.
>> Tell us more.
>> So, without going too
much into the technical depths of it,
I think it's taking a step back and looking at
where we are currently with deep networks,
and not forgetting that maybe
sometimes we've burrowed too far down
one rabbit hole and it might be time to come
back and look at a slightly different technique.
So, as we get into in the course,
we will see many times that letting
the machine learn end-to-end,
works really, really well.
And I think that capsules thing is kind of saying,
let's not bring too much innate knowledge to the problem,
let's try and learn it, let's try and teach
the machine the problem it needs to solve.
>> Fascinating. So, you giving
us a technological breakthrough,
and a soft skills slash,
overall business perspective and life perspective.
So, Ivan, Daire, when I was putting together
some questions to ask you today,
I came up with the idea of
there's a couple of few myths in
the space that I'd love to try and dispel with you.
And I came up with three, so,
if you'll entertain me here with some of your answers.
The first one is, the fact that we hear often, "Oh,
I can't get into this space
because you need lots of hardware and
computational power to do anything in
AI machine learning computer
vision". What are your thoughts Daire?
>> Well, what I think about that Andrew is,
part of what you'll see on this course,
is this course, is actually capable of pulling off
some pretty cool image segmentation.
These are real world applications that you can do,
and we're doing that without having the benefit of
huge GPU and Cloud clusters and things like that.
So, you'd be really surprised what you can pull off
without having access to GPUs and to CPU clusters.
Now, obviously if you have the access and you
feel going down that road, sure.
It's an absolutely fantastic resource to have.
But we find when we're working a lot of the time,
we do an awful lot of prototyping and stuff
like that locally on our own machines.
And it's only when we really want to get a real deep burn
we put it onto the GPU or
send it to the cloud for a big training.
>> Great. Well, I think in terms of getting started,
I want to consider that myth, busted.
The next one is kind of interesting.
The fact that people say you need
two Ph.D.s in Mathematics.
Okay, I'm exaggerating a little bit,
but that you have to be extremely proficient
in Mathematics to enter the space as well.
How do you feel about that one? Daire.
>> Again Andrew, I think this one is a complete myth.
>> Okay.
>> I mean, you do need
certainly a certain level of Mathematics,
but you definitely do not need a Ph.D.
in Mathematics to get active in this space.
There's a whole pile of activities that you can
do without understanding the underlying Math whatsoever.
>> Okay.
>> And then there's another bunch of things where knowing
the Mathematics is advantageous
to you, but not necessary.
So, you could go your entire career without ever
really delving into
the Mathematics underneath these algorithms.
Really, it's only people who
are developing improvements to
the algorithms themselves that really even there,
you don't need it.
>> Okay. So, is this
the difference between what we consider
the the thin layer of really hard core expert scientists,
versus the practitioners, versus other people? Ivan.
>> Yeah. Absolutely. I think
it is in the same way that you
don't have to be able to
make a car from scratch to drive one,
or you don't have to be able to
fabricate a computer processor to program a computer.
I think people can use libraries like
the Microsoft Cognitive Toolkit to very good effect,
and solve real world applications.
The deep Linear Algebra stuff
is useful in terms of understanding what's going on.
But as Daire mentioned, it's only really when
you're coming in to toolkits
like the Cognitive Toolkit and trying to
get the most performance out of it,
that you really need to be
totally au fait with the math.
I think otherwise, you can build
great apps just playing with these libraries.
>> Great. So, while I think that,
and it should be clear that all
Ph.D.s are welcome to even try this course,
and everybody's included in
terms of needing those Math degrees.
Myth, busted.
I think in this myth,
I'm going to hear, kind of
interesting answers, but I'll try it anyway.
And it says that, you need mountains of data to do
anything with accuracy in
the Computer Vision/AI/Machine Learning space.
Interested to hear your thoughts on that one, Ivan.
>> Yeah. Certainly there are many algorithms that needs
no data at all to train them.
Their just simple image processing.
As you move towards the deep networks,
you tend to need more and more data.
And it certainly used to be true
that large volumes of data were needed.
But since then, techniques like
Transfer Learning where you effectively take
knowledge from a pre-trained network for
a slightly different application,
that can reduce the volume
of data you need significantly.
The other thing you can do, tricks
like data augmentation.
So, for instance, if you were
trying to detect coffee cups,
and you had some sample photos of coffee cups,
you could just horizontally flip the images.
Now you've doubled your data. Tricks like these-.
Absolutely. So, you used to need huge amounts of data,
not so much anymore.
>> Okay.
Interesting. Where do you stand on this one Daire?
>> Yeah. Like Ivan says,
particularly with things like Transfer Learning,
you can leverage pre-trained models that
other people have done a whole bunch of work on,
and then you just do your software or doing
the rabbit detector or you just add a thousand images of
the rabbit to the top of
a pre-trained semantic segmenter
and it would be able to find rabbits for you.
>> Great. So, I think with some caveats that there are
times when you possibly need your own huge datasets,
with the techniques and technologies come a long way,
where that's no longer always the case.
And therefore, that's another myth we've just busted.
So, Ivan, Daire, what I want to do now is
do my Inside the Actor Studio type of question.
So, I just want to maybe go back in
time and imagine you've bump into your younger selves,
maybe just before Emdalo or even younger.
And what one piece of advice in terms of career,
and in terms of this space,
do you think you'd give your younger self that
would help them along their career?
It's very useful for the students I think to,
and the viewers to hear something like that.
So, Ivan, want to give that a try?
>> Sure. So, I think in
the field of Computer Vision and AI,
it can be quite daunting when you come to it first.
It's such a wide breadth of technologies,
and techniques, and math, and whatnot.
And, I'd say don't be afraid to start.
Start somewhere, you roll up
your sleeves and start working on the labs,
start building your knowledge, start trying stuff.
And as you do, incrementally,
your knowledge about the entire domain will improve.
I mean, that's the big thing
for me. Don't be afraid to start.
Start small and it'll come in time.
>> Great advice from a practitioner.
Thanks Ivan. Daire.
>> Yeah.
So, what I would say to someone
you know to my younger self,
for instance, if they were going to do
these courses is to persevere with the course.
Don't be one of these people who tries something,
and then tries something else,
and tries something else.
Do this course and persevere with it.
And persevering, it doesn't mean just
sticking to trying to solve it the same way.
There's a whole load of different ways
that different people solve problems.
For instance, for lots of people,
that's just not driving through
the problem itself but just to talk to the other people.
Particularly, we have forums organized.
Use those forums to
discuss the problem with other people.
Somebody else who thinks a little bit differently from
you here will be able to present you with a solution.
>> That's really good and
great advice to give your younger selves.
So, Ivan, Daire, that wraps
up our little chat here in the Treehouse.
And I think on behalf of the students,
but personally as well,
I just want to thank you both.
You came on this journey with me for this course.
And hopefully, the students
will really enjoy what you've put together.
It's an absolute pleasure to have
real world practitioners join me here. So, Ivan.
>> Thank you Andrew.
>> Thank you so much. Daire.
>> That's been great.
>> Thanks so much.
-------------------------------------------
7 things you didn't know about Artist -bombed - Duration: 6:56.
seven things you didn't know about artists attention to detail most artists
are introverts what is an introvert what is an introvert what is an introvert an
introvert is usually a person that stays to themselves so does that mean
artists are kind of shy and reclusive mostly (I'm not though) however do artists
love attention of course they do but we love getting attention through our artwork
you'll see when you meet your artist friends rolling on to number two
inspiration what keeps an artist going on and on and on you ask like an energizer bunny
inspiration for motivation there are so many people who make an impact on an
artist's life whether they know it or not and it's usually another artist any
type of art can inspire a young boy to grow up and still be creative and have
that love for creativity in his heart one of my favorite artists Mark Crilley
at the time be still on my favorite Artist I'll
leave a link after this video if you guys don't know who mark really is you can
go and check out his work it's amazing and he left a comment on one of my
deviantART posts which is an art website and that made me want to work harder
knowing that my talent is being seen by my idols rolling into number three do we
eat well no not really well unless you mean we eat and sleep
art and it can get a little bit unhealthy sometimes wow I really really
like this piece how that came out I really really got the lighting right
there it's just but when we get very involved into our work we really really
put everything that we are into our projects and that's another reason why
certain artists charge a lot of money when it comes up to certain projects a
lot of people don't know that a lot of work goes in to what we do especially as
our skills evolve into something even more amazing so we eat and sleep art
and we love it well this is number four we consider ourselves artists even when
other people don't consider an artist well what do you mean well what I mean
is sometimes certain art forms aren't really seen as art a good example of
this would be digital art some people think that the machines do all the work smh
so coming in at number five well us artists we hold a deep dark
secret that is we see all the beauty but we also see all the
all of the darkness in the darkness we see the shadows how the shadows contrast
with the light to make a beautiful contrast we see the colors orange green
blue yellow and we see them not like the normal eye but in a way that goes beyond we think we think
artists, other artists art is better than ours nine times out of ten so yeah
it's very motivating actually it makes us wanna work harder with our own art
and gives us new ideas to bring to the table and that's a lovely thing isn't it
I'm just about blank here what's number 7 what did I eat for breakfast today? what does
my future hold? could I become president one day? well a little off subject seven we are very very very very very
critical on ourselves yes we think all of our art is
basically
can use improvement so yeah wah wah wah yes if you ever see an
artist chewing up a piece of paper or something like that it's either because
he loved the piece of art and it's just delicious to him or he hated it so you
eats it up so yeah, yeah art delicious
hey I hope you guys enjoyed that video I put a lot of work into and it was fun
LRA-Creations wishing you peace happiness and love and as always let's
not forget art for without it there's no heart in the description or right here
you'll see Mark Crilley's page and you want to join the creation fam
subscribe hit the Notifications so you know when a new video drops and be well
-------------------------------------------
RUSSIAN GIRLS TALK ABOUT MOROCCO - Duration: 4:37.
For more infomation >> RUSSIAN GIRLS TALK ABOUT MOROCCO - Duration: 4:37. -------------------------------------------
Suspect In Gruesome Ventura Restaurant Stabbing Rants About Camera In Eye, FBI; Refuses Public Defen - Duration: 2:26.
For more infomation >> Suspect In Gruesome Ventura Restaurant Stabbing Rants About Camera In Eye, FBI; Refuses Public Defen - Duration: 2:26. -------------------------------------------
What students think about SAGE University, Indore - Duration: 0:46.
SAGE TALENT QUEST 2018
Không có nhận xét nào:
Đăng nhận xét