Using Neural Networks to Read Minds

December 16, 2019

Researchers from McGill University and the University of Montreal are decoding the brain using neural networks

In the development of artificial intelligence applications, the holy grail is the creation of an artificial neural network that functions like the human brain. This is an elusive goal, because the human brain is an extremely complex organ that functions in flexible and fluid ways that can be difficult to replicate in the world of AI.

Today, a team of researchers from McGill University and the University of Montreal are making breakthroughs with functional magnetic resonance imaging (fMRI) of people’s brains while carrying out various cognitive tasks. The goal is to develop better understand and create computational models of how the brain works, and then use those models to train artificial neural networks to map the images to actions quickly and accurately. Yes, we’re talking about mind reading.

This would be a big leap forward for the AI world, according to one of the lead researchers on the project, Dr. Pierre Bellec, an associate professor at the University of Montreal. Dr. Bellec is the scientific director of the Courtois Project on Neuronal Modelling (NeuroMod), which is spearheading the collaborative research effort.

“Something the brain does really well is to switch from one context to another,” Dr. Bellec explains. “It has very elaborate organization, and specialized networks and subnetworks, and those networks and subnetworks are able to reconfigure dynamically. By contrast, current architectures used by AI researchers are extremely specialized for certain types of tasks, and have a hard time generalizing over different contexts.”

The researchers hope that by mimicking the architecture of the human brain, they can develop a more versatile AI model that can generalize over different tasks, much the way the human brain does.

“AI has been an inspiration forever, but here we are not just drawing general principles, we are doing extensive imaging to map out the full activities of the brain in an unprecedented level of detail,” Dr. Bellec explains. “We’re hoping to be able to draw directly from rich data to gain insight on how the brain works, rather than drawing from general, vague principles.”

To collect the datasets for this ambitious effort, the research team has recruited a small group of volunteers — a half dozen subjects — to watch videos, look at images and play video games while they are in an MRI machine. The research team had to build a new game controller without any metal, printed in 3D plastic with a fiber optic cable connection! The machine allows the researchers to track and record the activity in the brains of the subjects as they carry out their tasks. The research team expects to gather many terabytes of data over the course of the five-year study, during which time each subject will spend around 500 hours in an MRI machine.

“Essentially, we are trying to find a new way to integrate activity from human neural networks to help train artificial networks,” Dr. Bellec says. “The hope is that if we manage to do that, we can create computational models of how the brain works. And potentially we can train new artificial neural networks that may perform better in some settings than what we have now.”

The computing infrastructure and initial benchmarks

To move this project forward, researchers from the University of Montreal teamed up with researchers from McGill University who have extensive experience in high performance computing and work with MRI images that require large memory capacities.

They also sought the help of Dell EMC and Intel, along with the supercomputing resources of the Dell EMC HPC and AI Innovation Lab in Austin, Texas. The team is using the lab’s Intel-based Zenith cluster, which includes hundreds of Dell EMC PowerEdge servers with Intel Xeon® Scalable Processors and the Intel® Omni‑Path Architecture.

“People have done a lot of training on GPUs,” Dr. Bellec says. “We wanted to use CPUs for this project, thinking a CPU architecture with a large memory capacity would be better for larger files. We thought this type of hardware would be a perfect fit for our use case, so we decided to run some benchmarks.”

After testing on a GPU architecture, the team found that a CPU-based model can maintain similar performance – with validation accuracy reaching 99% after 10 epochs for motor task, and 91% after 20 epochs for working-memory task – as the GPU based models, but requires much less training time – 20 minutes vs 3 hours per epoch using 10 CPUs and two GPUs, respectively. Considering that, in the real world, the CPU resources are more easily accessed and cost less, the project provides a more feasible solution for the application of deep neural networks on large-scale neuroimaging data by training the model directly on CPU hubs instead of waiting for the GPU resources.

For those initial benchmarks, which kicked off the NeuroMod project, the research team used publicly available datasets from the Human Connectome Project, according to team member Dr. Yu Zhang, a computational neuroscientist affiliated with the University of Montreal. The Connectome Project, which is mapping the neural connections in the brain, offers researchers access to fMRI scans from 1,200 subjects.

“Currently, we’re using that project data and trying to do the de-coding.” Dr. Yu Zhang says. “The neural networks take a short series of this fMRI data and try to predict the specific task the subject was performing during the scanning.”

In the benchmarks on the public dataset, the team is evaluating the performance of two different architectures — a traditional convolutional neural networks and a more complex one, called ResNet, that has been used a lot in image processing.

“The benchmarks we are doing are what we call brain decoding, or in layman’s terms, mind reading,” Dr. Bellec adds. “You look at brain images and try to guess what people were doing. In the next stage, we will have our own data, in which we play video games. The idea is to try to train an artificial neural network to play the video game in the style of the particular player.”

Big data and big memory

By its nature, NeuroMod is a project that needs an HPC system with a large memory capacity to handle terabytes of data. In the initial benchmarks on the public dataset from the Human Connectome Project, the research team was dealing with 9 terabytes of compressed data, or 20 terabytes in an uncompressed form.

“And that’s only for the brain images,” Dr. Bellec says. “We are also collecting data from all the videos, and we are collecting a lot of physical data on the subjects, including heartbeat, respiration and the small motion of their eyes in very high resolution. All those auxiliary data, which we will eventually use in the model, can add a tens of terabytes of data.”

This deluge of data makes it all the more important to have ready access to an HPC cluster with big memory, which is what the team is getting through the Dell EMC HPC and AI Innovation Lab.

“We got access to the cluster in June, and it’s been very productive for us,” Dr. Bellec says. “We have been able to run a number of benchmarks that we had not been able to run prior to that time. And this is just the beginning, or so we hope. We’re getting familiar with the computing hardware architecture for deep learning, which we plan to use for years to come.”

This is just part of the process when a research team is breaking new ground. Oftentimes, organizations need new HPC architectures that are built for the challenges of huge datasets and unique workloads.

“Many people are excited about being able to evolve neural networks in ways that are inspired by biology, and it’s increasingly clear that we need a different type of hardware to do that,” Dr. Bellec says. “And that’s what we have with the Zenith cluster in the Dell EMC HPC and AI Innovation Lab.”

To learn  more

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

Nvidia Appoints Andy Grant as EMEA Director of Supercomputing, Higher Education, and AI

March 22, 2024

Nvidia recently appointed Andy Grant as Director, Supercomputing, Higher Education, and AI for Europe, the Middle East, and Africa (EMEA). With over 25 years of high-performance computing (HPC) experience, Grant brings a Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Houston We Have a Solution: Addressing the HPC and Tech Talent Gap

March 15, 2024

Generations of Houstonian teachers, counselors, and parents have either worked in the aerospace industry or know people who do - the prospect of entering the fi Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire