An intelligent algorithm developed by a Canadian researcher could provide Tinder users with a better selection of potential matches by taking into account their 'swipe right' history.
Harm de Vries, a post-doctoral researcher at the University of Montreal, developed the algorithm after signing up to Tinder in Montreal in 2014. His user experience was not the best as he found himself swiping through hordes of pictures of women who were not his type.
“Tinder kept offering me photos of women with lots of tattoos and piercings, even though I'd never chosen a single one,” de Vries said. “I don't want to offend anyone, they're simply not my type.”
The engineer was surprised to learn that Tinder has no algorithm in place to pre-select potential matches based on the users’ past selection and only takes into consideration geographical proximity.
De Vries decided to fill the gap and created software that analyses the images of people who received a swipe to the right - a Tinder version of Facebook's 'Like' button - enabling every two mutually swiped-right users to start a conversation. To make the software reliable, de Vries extracted and processed almost 10,000 images from Tinder.
“Ten thousand images might seem like a lot, but in reality it was too few for the programme to be able to precisely predict which image might interest me, as physical attraction does not depend uniquely on objective characteristics such as hair colour,” de Vries said.
In order to establish his programme's success rate, de Vries' first had to figure out what his own preferences actually were.
“I realised that I was interested in 53 per cent of the women's portraits, which meant that my tastes are actually wider than I thought!” he said. The first version of the programme, which allowed the user to label images to train the machine, had a mixed result: 55 per cent.
“I labelled all 10,000 images from Tinder. 8,000 were used to train the program and the rest were used to evaluate the performance of the program. The results of the first version were hardly better than chance, because it seems that a sample of 10,000 photos was too little and because predicting attraction is more complex than a computer determining whether or not there's a person in the image,” he added.
To achieve better results, the engineer developed a system capable of so-called deep learning, which successively filters information from the photographs as well as labels to understand general characteristics.
Eventually, the program was able to determine with 68 per cent accuracy which women de Vries would like.
“A success rate of 68 per cent is a very good start. One of my good friends who knows my tastes well looked at a random sample and only achieved 76 per cent!” de Vries concluded.
The result leads de Vries to believe that artificial intelligence could improve computer analysis of Tinder users' preferences.