This has been a big few weeks for machine learning and imagery.(See links below). Now Facebook just published new research outlining a novel approach for generating realistic, artificial images of scenery and things like dogs, planes, deer, ships, trucks, horses, and, of course, cats.
I've tried parsing through the research, but the details are beyond me. What is interesting, however is their use of an approach called "Generative Adversarial Networks" (GAN). Essentially, what they've done is create a kind of feedback loop between two networks, where the first, the "generative network" generates an image from noise. Then the other, "discriminative network," takes that resulting image, and essentially compares it to training data that is based on real images (note: this is a slight simplification). The result is that with each iteration, the generative network is 'tricked' into generating increasingly realistic looking imagery.
The researchers then testing the resulting images with a group of volunteers and found that 40% of the images were realistic enough to fool a human into thinking they are real images.
What is Facebook likely to do with the results of this research? That's unclear, but with their Oculus Virtual Reality acquisition, it seems reasonable to assume that they are going to need cost-effective methods for generating a massive scale of virtual scenery and objects. Could this research represent early forays into that work?
Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
Looking Inside the Image Recognition of Artificial Intelligence:
Is This the First Computational Imagination?
#artificialintelligence #machinelearning #virtualreality #facebook
This beautiful illustration by walks you through the entire dialogue that the two modems are having: in this case, a pair of 19,200 baud modems* talking to each other and establishing how they can talk.
* One baud is one line level transition per second -- that is, the signal flips from 0 to 1 or vice-versa. That's almost but not quite the same as bits per second, because some of those transitions are used to ensure that the data made it across the (noisy) phone lines clearly. The most common configuration was "8-N-1," meaning that it would send eight bits of data, no parity bits, and one "stop" bit to indicate the end of a byte. (A parity bit is a type of checksum that the receiver can use to make sure no data was corrupted during transmission) So nine level transitions would be used to send eight bits of data, and so a 19,200-baud modem would send about 17,000 bits per second.
Compare that to your cable modem, whose speed is measured in megabits per second, and you realize how much has changed. But 19,200 was shockingly fast in its day: it meant that you could download a jpg in half a minute.
(Side note: When people use "mega" and "giga" for network speeds, they mean decimal mega and giga, i.e. one million and one billion. When they use it for RAM amounts, they mean binary mega and giga, i.e. 2^20 (1,048,576) and 2^30 (1,073,741,824). And when they use it for hard disk sizes, they mean "somewhere roughly in this region, because hard disk manufacturers lie like dogs about their capacities anyway." Confusion between these things has led to some pretty spectacular engineering problems over the years.)
h/t to , , and the various other people that have pointed me at this figure over the years.
- University of Maryland Baltimore CountyComputer Science, 1989 - 1990
- University of MumbaiComputer Engineering, 1984 - 1988
- St Xavier's College1982 - 1984
- St Joseph's High School1972 - 1982
- ABB Corporate ResearchSoftware Researcher, present
- S3 Technologies
- RWD Technologies
- Essential Computing Net
- Buzz (current)
Face It, Your Brain Is a Computer - The New York Times
For most neuroscientists, this is just a bad metaphor. But it’s still the most useful analogy that we have.
Pictures of Chinese People Scanning QR Codes — Chrysaora Weekly — Medium
On the way back to Shanghai from our fieldwork trip in rural Guizhou, Tricia and I decided to go shopping at Guiyang’s H…
Review: Pixar’s ‘Inside Out’ Finds the Joy in Sadness, and Vice Versa - ...
Pete Docter’s new animated film, with the voices of television comedians like Amy Poehler, Mindy Kaling and Bill Hader, brings to life an 11
Google Brain’s Co-Inventor Tells Why He’s Building Chinese Neural Networ...
Andrew Ng wants a truly intelligent Chinese language machine, and with Baidu he’s got the resources to build it
On Time-Lapse Rocket Ride to Trade Center’s Top, Glimpse of Doomed Tower
For less than four seconds, a jarringly familiar pinstripe facade will loom into view on one wall of the elevator cabs servicing the observa
Filter by Species All Trees American Linden Amur Maple Baldcypress ...
New York City's urban forest provides numerous environmental and social benefits, and street trees compose roughly one quarter of that canop
How do I remove bird's nests from my dryer ventilation?
Birds can choose the weirdest places to build nests. I'm sure Mama bird thought that your dryer vent was the perfect warm, cozy little spot
Motion sensing using the doppler effect Recently I stumbled upon ...
Motion sensing using the doppler effect. Recently I stumbled upon an interesting paper for implementing motion sensing requiring no special