Machine learning helps produce ‘wiring diagram’ of the human brain

MIT is using automated machine learning to help develop a ‘wiring diagram’ of the human brain.

Back in the 1980s, researchers at the University of Cambridge managed to map the neural connections of a tiny worm with just 302 neurons – after a dozen years’ work.

But such a diagram, or connectome, of the human brain could now be roughed out in as little as five years, using the new techniques.

The computer is given electron micrographs as well as human tracings of these images, and then searches for an algorithm that allows it to imitate human performance.

“Instead of specifying the details of how the computer does something, you give it an example of what you want it to do and an algorithm that tries to figure out how to do what you want,” says MIT‘s Viren Jain.

After the computer is trained on the human tracings, it is given  electron micrographs that have not been traced by humans.

The team has also invented new ways of evaluating how well the computer imitates humans at the task of tracing.

At first, it took months to come up with an accurate neuron-tracing algorithm. However, Jain’s team cut that time dramatically once they started using computers equipped with graphics processing cards, allowing them to perform computations 50 to 100 times faster. Now, it takes only days to produce a new tracing algorithm.

Some neuroscientists believe that mapping connectomes could have just as much impact as sequencing the human genome, by allowing researchers to discover which differences in the connectome are important in diseases like Alzheimer’s and schizophrenia.