Google’s Neural Network research


Google Research Blog: Inceptionism: Going Deeper into Neural Networks

One of the challenges of neural networks is understanding what exactly goes on at each layer. We know that after training, each layer progressively extracts higher and higher-level features of the image, until the final layer essentially makes a decision on what the image shows. For example, the first layer maybe looks for edges or corners. Intermediate layers interpret the basic features to look for overall shapes or components, like a door or a leaf. The final few layers assemble those into complete interpretations—these neurons activate in response to very complex things such as entire buildings or trees.

One way to visualize what goes on is to turn the network upside down and ask it to enhance an input image in such a way as to elicit a particular interpretation. Say you want to know what sort of image would result in “Banana.” Start with an image full of random noise, then gradually tweak the image towards what the neural net considers a banana (see related work in [1], [2], [3], [4]). By itself, that doesn’t work very well, but it does if we impose a prior constraint that the image should have similar statistics to natural images, such as neighboring pixels needing to be correlated.


“Children up to the age of seven had to wear a muzzle, which was provided by the Notional Health Service, and when they were old enough they were fitted with special safety dentures which were made from lead. So heavy were these dental implants that children found it virtually impossible to open their mouths, restricting them to only uttering a word or two.

The scheme backfired when a group of muted children naturally developed telepathic abilities and tricked the county’s dentists and orthodontists into boarding a mysterious black bus, which was never seen again.”

And more.


HT alicebartlett

A book about “gestural interaction in the digital everyday”


People walking around when talking on their mobile phone is a common behavior.
Referred to as “Cell Trance” in the Urban Dictionnary2, this way of moving
back and forth is often seen in public venues such as hallways, sidewalks, train
platforms, bus stops or shopping malls. To onlookers, the erratic perambulation
looks aimless, as if the caller is detached from his surroundings, absorbed in a
private sonic universe.

From Curious Rituals, a book about “gestural interaction in the digital everyday”, by Nicolas Nova, Katherine Miyake, Waton Chiu and Nancy Kwon.

via Alexis Madrigal and Jesper Balslev

The Merchandizing Origin Of Your Fake Way Of Getting Dressed For Work


The Merchandizing Origin Of Your Fake Way Of Getting Dressed For Work.

Ah, “Business Casual.” Do you know that it comes from this eight page brochure mailed out to about 25,000 corporate human resources types in 1992? A dude named Rick Miller and his PR team at Levi’s created this brochure to market Dockers through employers by creating an ersatz new dress “code” that employers could institute without thinking much.

And just like that, it worked.

You’re welcome.