Artificially Intelligent "Sorting Hat" Aims to Bring AI Image Classification Benefits to All

Designed to make the benefits of AI available to non-data scientists, this "Sorting Hat" works using Create ML — rather than magic.

Scientists at Okayama University, Hokkaido University, Universiti Tunku Abdul Rahman, The National Institute for Basic Biology, and the Nara Institute of Science and Technology have developed an AI-driven image classifier that they say can be readily tailored to a range of tasks — and they've likened it to a certain fictional school's magical Sorting Hat.

"Classifying images using AI usually requires a high level of computer knowledge," says Kiyotaka Nagaki, associate professor at Okayama University, of the team's work. "What we did is build AI models on a [Mac] computer with the Create ML app suitable for our own image samples. Moreover, the AI can be trained to become an order-made image classifier for any variety of images that suits one's purpose."

The team's own implementation of the system is designed for use by cell biologists, and the key to its success is that it's entirely usable by non-AI specialists thanks to the Create ML framework. Trained on chromosomal images, the system easily detects cells undergoing mitosis — and proved highly successful at sorting images into the right "houses" once trained.

"There are more trivial classifications in our lives than one might imagine. Automating such classifications by entrusting them to an AI can not only eliminate fluctuations caused by individual differences but also save many valuable research hours," Nagaki claims of the system, which is claimed to be easily picked up by non-data scientists. "Streamlining such trivial classifications make extensive image-based studies more reproducible and reliable."

"Since this system is inexpensive and can be easily trained via deep learning using scientists’ own samples," the team concludes of the work, "it can be used not only for chromosomal image analysis but also for analysis of other biology-related images."

The team's paper has been published in the journal Chromosome Research under closed-access terms, with the models available under the reciprocal GNU General Public License on GitHub — along with the ImageSorter image processing tool.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire:
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles