Chernoff faces showing multi-dimensional data in 2D: Google Quick Draw meets t-SNE, but from the 1970s. Via Tufte’s The Visual Display of Quantitative Information.
Google Translate Handwriting
Highlighting In Google Book
Old meets new: highlighting in a scan of a book via search term.
Auto-Translation of Norwegian
Google’s translation algorithms are kind of amazing, but it seems to have some trouble with Norwegian in this video. Here’s a sample from the first minute:
“Ministering from releasing also finished warnings from his dick saw why not and schloss good and what’s going on s on went who would be because the fun name didn’t know hood systemic was looking at Skynyrd immune from for their successful and move he said and ripples beings man comment what at some point hadn’t been on that phenomena height I wilson’s clear that the the docs ready to their allude to that of her ordered posts it comes to the house yes that is me…”
Turn on “closed captioning” for the entire translation.
Google Car Blur/Shadow
Google Street View car’s shadow/blur, from outside the AMC Lowe’s theater in Wayne, New Jersey.
Random Image, Similar Image
Random images made using Processing, auto-uploaded to my server, and used as a seed for a Google “similar images” search. Continue reading “Random Image, Similar Image”
The Internet = A Cat
As reported in Slate (that I somehow originally missed), when Google X (their secret research team) passed 10 million video stills into an algorithm, the “silicon cortex” formed the image of a cat, pictured above.
Via: Slate (and Wired, at some point)
js Poetry
Some JavaScript poetry from today’s Google Doodle source code:
|
1 |
else if (!a.La && (!a.Ka && !a.ha && !a.Ma) |
and…
|
1 |
Pa = "function" == typeof Ra ? Ra() : Ra; |
Via: Google (make it easier to read using jsbeautifier.org)
Google Image Search x2
Google image search for “Google image search”…
Google’s Ngram Viewer
Today I found Google’s Ngram viewer and raw data sets. Above is the frequency of the words “Jeffrey” and “Jeff” in all of Google’s scanned books from 1800-2000.
The query is insanely fast, considering the data that the viewer must have to read through. Also of note, the frequency is normalized to the number of books published that year.