I need this visualization of 1,372 vibrators in the form of a five-foot long painting to hang over my couch.
Creative technologist and co-founder of Creative Coding Amsterdam Sabrina Verhage made this t-SNE visualization by scraping images from four different sex toy websites. She used the ofxCcv library, which reads each image and labels it using a trained convolutional neural network, according to similarities in the images.
t-SNE, or t-distributed Stochastic Neighbor Embedding, is a machine learning algorithm that plots and visualizes that high-dimensional data. In this case, it's a rainbow of good vibes.
After participating in the Sex Tech Hackathon in 2016, Verhage told me via Twitter message, she got into sex tech hacking as a hobby. That inspired her to propose a vibrator hacking workshop at the NODE festival in Frankfurt next month.
"I had this idea, in order to open up the sex tech subject, to add a little sex to all of my work," she told me. "Then, recently, I've been catching up on the machine learning hype—learning some new skills—and consequently thought of doing a vibrator t-SNE."
Aside from an exercise in machine learning, Verhage said she was hoping to find aesthetic similarities in the toys, and that they might flow by color and form, creating a work of art. "But I also liked mapping the amount, and all the varieties, and showing how not all of them look like penises and not all of them are targeted to females either," she said.
Most of these designs she'd seen before. The rubber duck vibrator, however, was a surprise.
Subscribe to Science Solved It, Motherboard's new show about the greatest mysteries that were solved by science.