Design a black hole imaging AI trained on black hole simulations

How much of science is real now that rather than filling in the gaps with speculation and simulations, we are forming the building blocks on speculation and simulations. This is why the origins of man are re-imagined every year with a totally new theory. How much of climate science is pure computer models in which we don't even know most of the variables.?

Oh so that's how you meant it.
Yes, well. When you take measurements from things you generally do it from the direction the object is. I'm not going to take a photo of you by pointing my camera at the floor.

>hurr durr what is maths? We totally just RNG everything hurr durr

She's the lead programmer on the project, you dunce. She could be talking on a street corner and it wouldn't change what's she saying.

The point is that if you feed an AI random data and it produces a black hole, then feeding it random data from space will make it product an image of a black hole. It's not surprising, and it's not proof of a black hole.

It looks like one of my eclipse photos, but worse.

Attached: 20170821_130924.jpg (5312x2988, 2.22M)

>maths
kek

Now zoom in to that pic 500% and make a news article about how you found the black hole

I think he's basically saying its how if you train some algorithm to output 0-9, when you give it random data it will confidently give you 0-9 because it doesn't know anything else.

>hurr me don't understand basic maths is very hard hurr me good programmer and physicist durr

Attached: 1554938255155.jpg (700x700, 209K)

So, it doesn't chance the fact she is discrediting you? Like literally stating they didn't train it on black hole simulations because that could result in a biased result.

It's not random data, though, any more than this image file is random data. Depending how you look at the data in this image it could be incomprehensible garbage. Or it could be Freddy Mercury singing.

Attached: Queen - Somebody To Love (Official Video)-kijpcUv-b8M 0001.png (616x360, 212K)

Then he's an idiot because he cherry picked some line from the talk and completely missed the section where she says they didn't train the algorithm to spit out pictures of black holes because then it would be biased.

photoshop

Congrats, first new black hole meme

Attached: emc2wtfomg.png (976x549, 228K)

Kek

Science degenerated into pop sci sensationalism long time ago. Higgs boson discovery, now this computer generated photo.
They do it to legitimate shit ton of money throwed at their useless projects

Attached: 1554938500613.png (1562x910, 436K)

I think I get what OP is trying to say.
Is the pic computer generated or an actual photo?

>Is the pic computer generated or an actual photo?
That depends on your definition.
Do you consider photos taken on your cellphone to be computer generated or an actual photo?
The same answer applies to this.

Not bad, not bad

But it makes sense.

what even is a photograph

What goes through the minds of people like this? What are they trying to achieve?

>hurr our eyes aren't real
Is the image a photography of the black hole or was generated from zero with no external visual data?

skip this and go directly to 6:44
>there are infinite images that are perfectly consistent with our telescope measurements
which is what OP says.
...and that's a great catch OP. hats off muh dude.
they picked the prettiest image to show, not the most accurate, because they don't have the most accurate, they have a shitload of potential images, they picked the prettiest... and what's the prettiest? the one that was closer to the fake black hole we saw in a movie.

So what would it take to please you, OP
an actual visible light color photograph?

Ever seen a photo of the sun? guess what thats not a photo either. doesn't mean its all made up bullshit.

>Is the image a photography of the black hole
This.

When you take a photo with your cellphone, you could generate an infinite number of images that would be consistent with the data.
I'm sorry, but you have no idea about digital imaging.

Its as if you just ignored the part where she talks specifically about how they didn't train the AI on simulated black holes because it would introduce bias.

>the one that was closer to the fake black hole we saw in a movie.
What "real" image were you expecting exactly?

Popular science has largely become a faith-based religion / hobby / entertainment source. Things don't need to be scrutinized, challenged, or even verified by third parties prior to mass distribution. Whatever "they" say is automatically trusted by the average reader. And even if it wasn't trusted, good luck reproducing or verifying the findings of projects on this scale yourself. You have no choice but to trust that the man on the pulpit is not only competent, but also reporting honestly. Otherwise you're a tinfoil hatter.

What's their intent? At the surface, to continue receiving funding and to continue selling clicks for the parties that approve that funding. The deeper conspiracy says that they are misdirecting people from researching the actual mechanism of gravity (which is currently not even close to understood), which could potentially lead to a totally different model of astrophysics being built. But I don't know about that shit.

Maybe its just arrogance and confirmation bias driving data interpretation. Who knows. Our best scientists used to think the Earth was flat. I wonder how hard they fought to maintain their position of authority?

what movie was this? the only black hole I can remember seeing in a movie was in Interstellar, and it sure as fuck didn't look like these AI generated pictures.

Attached: interstellar-black-hole-scene[1].jpg (2560x1440, 214K)

couldn't they just run the same program on light from some random direction to see if they get the same outcome?

If you watch from about the 12th minute, that's basically what they did. Using the same algorithm they fed it data from other photos, stock photos and so on, and it built an image that closely matched the original photo.

Wait

they didnt actually image the black hole but took instrument data and reconstructed what it could look like?

Attached: 1554506822986.jpg (768x768, 93K)

I don't think they claimed to have taken a photo but released an "image" of the messier 87 super massive blackhole.

>an image generated from externally gathered data isn't a photograph

just what do y'all think a photograph is, exactly? do you think because large area photos are taken from multiple stitched together photos, often using AI, aren't technically photos?

OP is just being contrarian.

>but took instrument data and reconstructed what it could look like?
Welcome to digital photography.

The vast majority of astronomy is all about fitting models to limited observations.

spectrometry is a thing that exists, and black holes have a completely light different signature than an eclipse or planetary transit.

To be fair, they cross compared across three independent data sources and they all supported each other.
The papers are free to read user.

Yea but they never even took a photograph...

heavy.com/news/2019/04/katie-bouman/

>We developed ways to generate synthetic data and used different algorithms and tested blindly to see if we can recover an image. We didn’t want to just develop one algorithm. We wanted to develop many different algorithms that all have different assumptions built into them.

They literally said I think it should look like this so lets make an algorithm they will make it look like this....

>they took the “sparse and noisy data” from a series of telescopes to make an image, according to CNN

You can write a piece of code to make any image from any data set you want. The only reason its news is because a female worked on the project

Attached: 1482955049146.jpg (798x612, 146K)

>When you take a photo with your cellphone, you could generate an infinite number of images that would be consistent with the data.
>I'm sorry, but you have no idea about digital imaging.
we are not even talking about visible light here idiot.
we do not talk about correcting sensor hue, or denoising the data and what not.
we are talking about data beyond the visible spectrum.

exactly, "real".
they are passing this as the actual picture of black hole.
when they produce a photo of a nebula, they state that this is artificial colouring as they take data from different spectrums and match them to rgb colours.
the difference with nebulas is that the shapes are real, the colours are fake.
this picture was chosen among many results that didn't even have circles, or holes or light halos.
interstellar.

Actual photo

Attached: 933_BIG_P_COLOR_2_TRUE_COLOR1_1980.jpg (1300x1300, 907K)

Is that Pluto?

This is the problem
The image was never taken it was generated from a data set
I would be fine if they used that to process and correct the light from infrared or any other non visible to visible so we can see what it looks like.
Problem here is this just a blurry ass CGI render based on "data" which was part of 100s of other images most of which were probably completely fucked beyond recognition

Dont fucking announce youve photographed a black hole but its a damn CGI render based on some murky ass data put through an algorithm

Attached: print.jpg (2400x3000, 3.74M)

it looks like they're trying to summon ghidorah

this is not difital photography phaggot.
this is not fucking digital image processing.
god damn it, how hard is it to understand that they don't have lenses and they don't place sensor elements to pixels on a file.

here's what they did.
it's as if you take 6 random photos at 6 different directions at the sky and scramble the data in a way that it might look like a cloud. you will get scrambled data that look like dicks, white noise, assholes and you'll get 1 that looks like a cloud.
then you chose that one to keep as "real" because this is what fits your rhetoric.
if there is no linear correlation to the actual data and the result, you can't claim it as "real" photograph.
I can get pi digits and feed them to a tiff file and someday it will produce something that might look black, and round, with curly hair around it.

Wasn't criticizing or denying anything, just stating how it was. There's no CSI toolkit that world governments use to enhance these nondescript blobs behind the scenes, extrasolar astronomy is shitloads of model fitting and speculation based on direct and indirect observations, and that's a lot of what makes it so interesting in the first place.
>They literally said I think it should look like this
Not really, the underlying theory behind black holes isn't random shit people pulled out of their asses for social media brownie points. We may not have seen a black hole, but with the decades of aforementioned modelling we still had a pretty solid idea of what we should expect to find.
I don't see your point. Their model fit a few different possible images, they chose the most aesthetically pleasing one to show to and hopefully excite the public. There's nothing particularly malicious, dishonest or damaging about that.
>they take data from different spectrums and match them to rgb colours.
If you don't even know what the visible spectrum is, I really don't think you're in any place to grandstand about authenticity in astronomy.

>There's nothing particularly malicious, dishonest or damaging about blatantly lying in order to get more grant money

that's the point I am trying to make.
the infamous clip of the objects moving around the S. A* is 1-on-1 correlated with the actual gif that was posted on a thread today.
they read gamma rays from the center of the milky way, which are high energy photons, they mapped them to gray scale according to their readings and produced that image.

>what we should expect to find.
we have no fucking idea what it would actually look like...
This is still an assumed image based on what we think it should look like. Not what it actually looks like
Its one of 100s of fucked up images but its the one that looks the most right to us. It should not be pushed as the first image of a black hole because thats not even close to true. We had amazing CGI in interstellar seen here which is exactly the same as the blurry ass image released
Its based on data of what we think it should look like

>Yea but they never even took a photograph...
They collected measurements. Exactly what the image sensor on your digital camera does.
Yeah, they didn't expose a film to light.

Measurements could mean anything.

>we are not even talking about visible light here idiot.
>we do not talk about correcting sensor hue, or denoising the data and what not.
>we are talking about data beyond the visible spectrum.
What is your point? You think that digital imaging is only visible spectrums?

>electron microscopy isn't real pictures so bacteria are fake news

What's your alternate theory then? How would you interpret the data?

WHY WLOULD A COMPANY PAID BILLIONS LIE ABOUT THERE SUCCESS? HMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMI DO NOT UNDERSTAND WHY.

True.
How about another example. If I use LIDAR to create an "image" of the interior of my room, is it "real" or is it CGI?

What company are you talking about here?

But they're not lying, that's the point. They took real data and fit it to a model, and then took the nicest looking fit to parade around, yes, maybe even to excite people with something a little bit more relatable and get more funding. I'm sorry you're so bitter about it.
>we have no fucking idea what it would actually look like...
Black holes themselves are still poorly understood, but the superficial aspects of their appearance? Not really. It's a dense massive object with a disk. This shit doesn't change.
> We had amazing CGI in interstellar seen here which is exactly the same as the blurry ass image released
It's almost as if the CGI in Interstellar actually had some scientific basis. Really makes you think, doesn't it?
>Its based on data of what we think it should look like
Man, once again, you can't even grasp basic astronomical terms and you're trying to tell us what we *think* things are like. We may not see things very clearly, but we aren't pulling everything out of our asses, either.

NASA

NASA had nothing to do with this.

>muh nasa boogyman

so if its not based on simulations and its a real picture why did they need a supercomputer to create the image?

>we aren't pulling everything out of our asses, either.
we pulling everything from math models.
that's what happen when you predate materialism

>we pulling everything from math models.
Which isn't an ass, regardless of whether you personally understand it or not.

Because they had 5PB of data to sift through. Imagine for a second you want to take a photo of a building 10 miles away with your cellphone. It only takes up a very small portion of the entire frame so you end up with a lot of data that's NOT the building. This is EVEN WORSE. Like taking a picture of an ant that is on that building.

What information do you have? Any?

Or they don't know what they are doing and just making up shit. If they need that much data, I am leaning towards they have no idea what they are doing.

I have some knowledge of technology.

>If they need that much data, I am leaning towards they have no idea what they are doing.
If you think that, I'm leaning towards you have no fucking clue.

>it's fake because I don't understand the data
haha got em how will those kike womyn ever recover

Not what I am asking.

Shut up you nazis, it just is.

Where in your butt does it hurt?

i'm literally prolapsing right now trying to hold back this shit so i can take it on your mom's chest later

You literally have no idea what are radio waves, laser and interferometry. All you need is to point at the right place and see the needle oscillate. That's all the data you need for everything that is real. But you are too dumb to understand the oscillation.

Physics have been selling bigger crap than this like big bang, space-time, expanding universe, string theory etc. Doesn't need to make sense, as long as something can excite enough normies, their grant money will continue to flow.

This is evil

The roast just threw together some python packages to generate a nice looking image for the public. The real research and data behind it hasn't changed, and it still holds true regardless if you think the image is dishonest or not.

Attached: paper.png (562x325, 73K)

What ARE you asking? I have seen the press releases, some of the presentations. I have some knowledge of digital imaging. The biggest sticking point here is the definition of "photograph" or "image". "The data can be interpreted to generate an infinite number of images that would match the data", so could any RAW data that comes from an image sensor! We have decided a way to interpret that data in such that when it is presented on the screen it appears like what we saw with our eyes. But that's still an interpretation of the RAW data that was collected. It's not just straight data in data out there's your picture.

>implying space is real
>implying earth isn't flat
Just fuck off.

OK, so how do you GUARANTEE that this image is real and no pulled from their asses given what n would happen if they didn't get a descent mage after a few years and needed to fund other useless projects?

It never hurts my butt to argue with retards like you.

give me another racewar please

That's the point: it uses the unreliable Machine Learning approach that works well mostly with supervised learning but terrible at guessing by itself.

>big bang
we can look back as far as we can and see dense plasma clouds, then a recombination period where protons and electrons combined to shoot off a shit ton of photons and making the CMB. Anything before that is conjecture.
>space time
it does appear to exist, perhaps not exactly to our current understanding, but the effects are still real.
>expanding universe
red shift?
>string theory
that's pretty much a meme, just math autists jacking themselves off

Said with all the authority of someone with zero qualifications.

>1 trillion dollars
>a fucking orange circle
Could have just opened up ms paint and saved everyone the effort.

Attached: black hole.png (787x518, 3K)