Science-is-magical - Science Is Magic

Paint colors designed by neural network, Part 2

So it turns out you can train a neural network to generate paint colors if you give it a list of 7,700 Sherwin-Williams paint colors as input. How a neural network basically works is it looks at a set of data - in this case, a long list of Sherwin-Williams paint color names and RGB (red, green, blue) numbers that represent the color - and it tries to form its own rules about how to generate more data like it. 

Last time I reported results that were, well… mixed. The neural network produced colors, all right, but it hadn’t gotten the hang of producing appealing names to go with them - instead producing names like Rose Hork, Stanky Bean, and Turdly. It also had trouble matching names to colors, and would often produce an “Ice Gray” that was a mustard yellow, for example, or a “Ferry Purple” that was decidedly brown.  

These were not great names.

There are lots of things that affect how well the algorithm does, however.

One simple change turns out to be the “temperature” (think: creativity) variable, which adjusts whether the neural network always picks the most likely next character as it’s generating text, or whether it will go with something farther down the list. I had the temperature originally set pretty high, but it turns out that when I turn it down ever so slightly, the algorithm does a lot better. Not only do the names better match the colors, but it begins to reproduce color gradients that must have been in the original dataset all along. Colors tend to be grouped together in these gradients, so it shifts gradually from greens to browns to blues to yellows, etc. and does eventually cover the rainbow, not just beige.

Apparently it was trying to give me better results, but I kept screwing it up.

Raw output from RGB neural net, now less-annoyed by my temperature setting

People also sent in suggestions on how to improve the algorithm. One of the most-frequent was to try a different way of representing color - it turns out that RGB (with a single color represented by the amount of Red, Green, and Blue in it) isn’t very well matched to the way human eyes perceive color.

These are some results from a different color representation, known as HSV. In HSV representation, a single color is represented by three numbers like in RGB, but this time they stand for Hue, Saturation, and Value. You can think of the Hue number as representing the color, Saturation as representing how intense (vs gray) the color is, and Value as representing the brightness. Other than the way of representing the color, everything else about the dataset and the neural network are the same. (char-rnn, 512 neurons and 2 layers, dropout 0.8, 50 epochs)

Raw output from HSV neural net:

And here are some results from a third color representation, known as LAB. In this color space, the first number stands for lightness, the second number stands for the amount of green vs red, and the third number stands for the the amount of blue vs yellow.

Raw output from LAB neural net:

It turns out that the color representation doesn’t make a very big difference in how good the results are (at least as far as I can tell with my very simple experiment). RGB seems to be surprisingly the best able to reproduce the gradients from the original dataset - maybe it’s more resistant to disruption when the temperature setting introduces randomness.

And the color names are pretty bad, no matter how the colors themselves are represented.

However, a blog reader compiled this dataset, which has paint colors from other companies such as Behr and Benjamin Moore, as well as a bunch of user-submitted colors from a big XKCD survey. He also changed all the names to lowercase, so the neural network wouldn’t have to learn two versions of each letter.

And the results were… surprisingly good. Pretty much every name was a plausible match to its color (even if it wasn’t a plausible color you’d find in the paint store). The answer seems to be, as it often is for neural networks: more data.

Raw output using The Big RGB Dataset:

I leave you with the Hall of Fame:

RGB:

HSV:

LAB:

Big RGB dataset:

Blowing Bubbles in the Gamma-ray Sky

Did you know our Milky Way galaxy is blowing bubbles? Two of them, each 25,000 light-years tall! They extend above and below the disk of the galaxy, like the two halves of an hourglass. We can’t see them with our own eyes because they’re only apparent in gamma-ray light, the highest-energy light in the universe.

We didn’t even know these humongous structures were smack in the middle of our galaxy until 2010. Scientists found them when they analyzed the first two years of data from NASA’s Fermi Gamma-ray Space Telescope. They dubbed them the “Fermi bubbles” and found that in addition to being really big and spread out, they seem to have well-defined edges. The bubbles’ shape and the light they give off led scientists to think they were created by a rapid release of energy. But by what? And when?

One possible explanation is that they could be leftovers from the last big meal eaten by the supermassive black hole at the center of our galaxy. This monster is more than 4 million times the mass of our own Sun. Scientists think it may have slurped up a big cloud of hydrogen between 6 and 9 million years ago and then burped jets of hot gas that we see in gamma rays and X-rays.

Another possible explanation is that the bubbles could be the remains of star formation. There are massive clusters of stars at very the center of the Milky Way — sometimes the stars are so closely packed they’re a million times more dense than in the outer suburb of the galaxy where we live. If there was a burst of star formation in this area a few million years ago, it could have created the surge of gas needed to in turn create the Fermi bubbles.

It took us until 2010 to see these Fermi bubbles because the sky is filled with a fog of other gamma rays that can obscure our view. This fog is created when particles moving near light speed bump into gas, dust, and light in the Milky Way. These collisions produce gamma rays, and scientists had to factor out the fog to unveil the bubbles.

Scientists continue to study the possible causes of these massive bubbles using the 10 years of data Fermi has collected so far. Fermi has also made many other exciting discoveries — like the the collision of superdense neutron stars and the nature of space-time. Learn more about Fermi and how we’ve been celebrating its first decade in space.

Make sure to follow us on Tumblr for your regular dose of space: http://nasa.tumblr.com

Uhmm, how exactly were all of those megafauna able to grow that large and function??? And how the fuck was that giant bird actually able to fly????????

Realistic answer?  Mostly because humans hadn’t come around and hunted them all to extinction yet.  Dinosaurs are exempt from this because they vastly predated us, but almost anything that coincided with our timeline, we killed.

We are, for our relatively small size and frail, sometimes clumsy physical characteristics, a TERRIFYING species.

Evolution has produced all kinds of Big Shit.  Ever seen a Paraceratherium?

Or a size chart for Sauropods that wasn’t produced before 1970?

Evolution likes to make things big.  It tries this all the time.  Whenever there’s a plentiful food source and enough space, things just get bigger and bigger.  

The largest animal to have ever lived is alive right now.  It’s the blue whale.  And it’s truly a masterpiece of evolution’s drive to Go Bigger.

HOW THIS HAPPEN.

Basically, because they live in the ocean, space isn’t really an issue for them, and thanks to buoyancy, neither is their frankly ALARMING weight.  The only real limit to their size is chemistry – whether they can possibly metabolize enough energy fast enough to stay alive at their size.  Blue whales are estimated (having, for obvious reasons, never been measured in one piece) to be able to reach over 200 tons.  As an average weight.  Fluctuating with their feeding season.  This was for a 98 foot long whale.  The longest whales ever measured were 110 feet and 109 feet, both females.  (Males tend to be slightly shorter, but heavier at any given length).

A blue whale can hold over 90 tons of food and water in its mouth.

They need 1.5 million kilocalories of food per day.

Blue whales are MASSIVE.

They are not the LONGEST animal in the world, though, just the heaviest.  The longest is likely one of two things: the Lions Mane Jellyfish or the Bootlace Worm.  The longest recorded Lions Mane Jellyfish washed up on shore with tentacles measuring 120 feet.  It is unknown if they can be longer than this, but certainly possible given how fragile they are and the fact that this is just one that happened to get washed up on a beach.

The longest recorded bootlace worm SMASHES this record, but because of its stretchy body and the date of the recording (1864), the scientific accuracy is disputed.  It also washed up on shore and measured 180 feet.

How far off topic am I this time?

Anyway yes animals get big sometimes.  It helps deter predators when you’re too big to be hunted by anything.  The only natural predator of the blue whale is killer whales.  Regarding bears specifically, Brown Bears are in far more trouble than Black Bears because the Brown Bear line trends toward going bigger, which makes them easier targets for humans, while Black Bears have evolved to be shy and stealthy and avoid human contact.  As a result, Brown bears are far larger, but far more likely to be driven to extinction by humans.  Nature functions just fine if it’s left alone.  We just ruin everything we touch.  That’s why the largest individual crocodiles still living right now are the ones that have learned to avoid humans at all costs: conservation laws have not protected crocodiles from poaching long enough for them to get really, really big, even though we have significant historical records of crocodiles larger than what we generally see now.  At least some of those records are considered to be reliable and put a couple extant crocodile species well over 20 feet – some over 22.  The largest reliably measured crocodile was Lolong, a Saltwater crocodile in the Philippines who measured 20 feet 3 inches and died a few years ago.

And Argentavis magnificens was able to fly because it was designed to.  Even with a massive 24 foot wingspan, it only weighed around 175 pounds, because birds have very lightweight skeletons.  As impressive as the size was,

a living bird of that size probably weighed about as much, if not a bit less, than the man standing next to it.  The surface area of its wings would have been sufficient to keep it in the air, mostly by gliding the way you see large modern birds of prey do.  It would have resembled a condor or vulture, just much larger.

A physical visualization of a magnetic field.

(Source)

hey guys i think i got a pretty nice tan over the summer, what do you think?

before:

after:

Drone with grabbing claw arms can lift 44 pounds

Prodrone’s latest creation could lift a four-year-old child, and uses its 5-axis metal claws to perch on fences like a bird.

This terrifying eel-robot will perform maintenance on undersea equipment

Nope.

Here’s how nightmarish humans would look if our bodies were designed to survive car crashes

Article by Chris Weller, Tech Insider & Business Insider

If you’re ever in a car with Graham, then don’t bother telling him to buckle his seat belt. His body is already designed to withstand high-speed impacts.

Designed by a trauma surgeon, an artist, and a crash investigator, Graham is a hypothetical scenario come to life. Supported by Australia’s Transport Accident Commission, the project is meant to highlight how vulnerable humans are to injury.

Graham, however, is not.

Keep reading

cyanobacterium: i have made Oxygen

chemotrophs: you fucked up a perfectly good planet is what you did. look at it. it’s all rusty

The folks over at NASA just featured this nifty infographic on APOD about detecting objects in the sky:

How to Identify that Light in the Sky

What is that light in the sky?

Perhaps one of humanity’s more common questions, an answer may result from a few quick observations.

Image: HK (The League of Lost Causes)

For example — is it moving or blinking? If so, and if you live near a city, the answer is typically an airplane, since planes are so numerous and so few stars and satellites are bright enough to be seen over the din of artificial city lights.

If not, and if you live far from a city, that bright light is likely a planet such as Venus or Mars — the former of which is constrained to appear near the horizon just before dawn or after dusk.

Sometimes the low apparent motion of a distant airplane near the horizon makes it hard to tell from a bright planet, but even this can usually be discerned by the plane’s motion over a few minutes. Still unsure?

The above chart gives a sometimes-humorous but mostly-accurate assessment. Dedicated sky enthusiasts will likely note — and are encouraged to provide — polite corrections.