Brain Decoding: Using Neural Networks to Read Minds

In this slidecast, Luke Wilson from Dell EMC describes a case study with McGill University using neural networks to read minds.

Developing a fuller understanding of the brain’s circuitry represents a huge challenge, with an estimated 100 billion neurons in the brain and 1000 trillion neuronal connections (10,000 per neuron) triggering thoughts and actions. It’s also a key step toward translating neuroscience research into clinical practice and public health in ways that will improve brain health in Canada and around the world.

"In this project, McGill University was running into bottlenecks using neural networks to reverse-map fMRI images. The team from the Dell EMC HPC & AI Innovation Lab was able to tune the code to run solely on Intel Xeon Scalable processors, rather than porting to the university's scarce GPU accelerators."

and

[avibank]

Why not do preprocessing on the GPUs? Maybe using new V100s with 32 gig vRAM might change the approach. Having said that, I totally get using cpus for the reasons Luke explained. Very cool.

[avibank]

Always wanted to do some similar research. This is a good step in the right direction!