Researchers aim to replicate human learning in a machine

learning-machineThe last year or so has seen a number of significant advances in our ability to build machines capable of learning for themselves.  For instance, we’ve developed machines capable of appreciating stories, understanding humor, understanding what they ‘see‘, detect sarcasm and even grasp empathy.

As well as advancing the machines ability to learn new things, there has also been progress made in how machines learn.  For instance, a team from Brown University developed robots capable of learning from each other.

To continue this work, a Harvard based team have been awarded $28 million to develop machine learning algorithms that push the frontiers of neuroscience.

Pushing the boundaries

The funding comes from the Intelligence Advanced Research Projects Activity (IARPA), which is designed to support work exploring the toughest challenges facing AI today.

The aim is to better understand why humans are so good at learning, and then design computers to replicate that.  They will try and achieve this by recording what the visual cortex does in minute detail, creating a map of its connections and then reverse engineering this to inform the next generation of algorithm.

“This is a moonshot challenge, akin to the Human Genome Project in scope,” the team say. “The scientific value of recording the activity of so many neurons and mapping their connections alone is enormous, but that is only the first half of the project. As we figure out the fundamental principles governing how the brain learns, it’s not hard to imagine that we’ll eventually be able to design computer systems that can match, or even outperform, humans.”

The team believe that if they’re successful, such systems could help to power everything from MRI scanners to driverless cars.

In the beginning

The researchers begin their work by training rates to recognize various visual objects on a screen.  As they learn, their brain activity is monitored in minute detail using the latest laser microscopes that were developed specifically for this project.  They allow the team to see how the brain works at a neuron level.

A section of the rats brain will then be taken to a lab where it will be put under a multi-beam scanning electron microscope (the first in the world).

“This is an amazing opportunity to see all the intricate details of a full piece of cerebral cortex,” the team say. “We are very excited to get started but have no illusions that this will be easy.”

Enormous data

This process will produce over a petabyte of data, with algorithms then going to work on it to reconstruct cell boundaries, synapses and connections, before visualizing them in three dimensions.

“This project is not only pushing the boundaries of brain science, it is also pushing the boundaries of what is possible in computer science,” the team reveal. “We will reconstruct neural circuits at an unprecedented scale from petabytes of structural and functional data. This requires us to make new advances in data management, high-performance computing, computer vision, and network analysis.”

This ambitious project would already be pushing the boundaries if it stopped there, but the team hope to use the understanding of the brain this 3D image provides them to then learn how the brain uses these connections to process information so effectively.

Nearly all AI projects are trained on huge dumps of data, so if we can improve the speed by which machines learn it would be a sizable move forwards.

They’ll do this by building algorithms for the kind of pattern recognition we do so well in the hope that it will enable machines to learn after a couple of exposures to an object rather than the thousands that are currently required.

“We have a huge task ahead of us in this project, but at the end of the day, this research will help us understand what is special about our brains,” the team conclude. “One of the most exciting things about this project is that we are working on one of the great remaining achievements for human knowledge — understanding how the brain works at a fundamental level.”

If you liked this post, buy me a coffee


Leave a Reply

Your email address will not be published. Required fields are marked *