Brian Kenji Iwana and Seiichi Uchida at Kyushu University in Japan set out to see if a machine is capable of classifying books in the same way as people, by looking at their covers.
The scientists downloaded 137,788 unique book covers from Amazon.com, from 20 different genres. Using the data, they trained a Convolutional Neural Network (CNN) to predict the content of a book based on its cover.
CNN is a type of artificial neural network, a computer model designed to simulate the way in which the brains of humans and other living creatures process information.
The researchers designed a CNN with four layers of up to 512 neurons each, which work together to recognize the connection between the cover design and genre.
Out of 20 possible genres, the computer program listed the correct genre in its top three choices over 40 percent of the time, and found the exact genre more than 20 percent of the time.
"This shows that classification of book cover designs is possible, although a very difficult task," the researchers told MIT Technology Review.
The machine found categories such as "computers and technology" and "travel" particularly easy to classify, while biographies, memoirs and books about politics and the social sciences were more difficult, the researchers said in their paper, published on Arxiv.org.
However the scientists did not compare the performance of the CNN to that of humans performing the same task, making it thus far impossible to know if machines are better than humans are at judging a book by its cover.
MIT noted that the research could help designers improve their skills when it comes to book covers.
"A more likely outcome, however, is that it could be used to train machines to design book covers without the need for human input," MIT wrote.
"And that means book cover design is just another job that is set to be consigned to the history books."