System Bits: July 24

Computers that mimic the human brain
According to a group of researchers led by the Jülich Research Centre in Germany, a computer built to mimic the brain’s neural networks produces similar results to that of the best brain-simulation supercomputer software currently used for neural-signaling research.

The custom-built computer named SpiNNaker, which the team said has been tested for accuracy, speed and energy efficiency, is believed to have the potential to overcome the speed and power consumption problems of conventional supercomputers, with the aim of advancing the knowledge of neural processing in the brain, to include learning and disorders such as epilepsy and Alzheimer’s disease.

Dr. Sacha van Albada, lead author of the study and leader of the Theoretical Neuroanatomy group at the Jülich Research Centre said, “SpiNNaker can support detailed biological models of the cortex—the outer layer of the brain that receives and processes information from the senses—delivering results very similar to those from an equivalent supercomputer software simulation. The ability to run large-scale detailed neural networks quickly and at low power consumption will advance robotics research and facilitate studies on learning and brain disorders.”

Dr. Sacha van Albada, leader of the Theoretical Neuroanatomy group at the Jülich Research Centre, Germany.
Source: Jülich Research Centre

The team reminded that the human brain is extremely complex, comprising 100 billion interconnected brain cells. While it is understood how individual neurons and their components behave and communicate with each other and on the larger scale, which areas of the brain are used for sensory perception, action and cognition, less is known about the translation of neural activity into behavior, such as turning thought into muscle movement. They said supercomputer software has helped by simulating the exchange of signals between neurons, but even the best software run on the fastest supercomputers to date can only simulate 1% of the human brain.

Further, it is presently unclear which computer architecture is best suited to study whole-brain networks efficiently. The European Human Brain Project and Jülich Research Centre have performed extensive research to identify the best strategy for this highly complex problem. Today’s supercomputers require several minutes to simulate one second of real time, so studies on processes like learning, which take hours and days in real time are currently out of reach.

There is still a huge gap between the energy consumption of the brain and today’s supercomputers, but neuromorphic (brain-inspired) computing allows for the investigation of how close electronics can get to the energy efficiency of the brain.

Developed over the past 15 years and based on the structure and function of the human brain, SpiNNaker — part of the Neuromorphic Computing Platform of the Human Brain Project — is a custom-built computer composed of half a million of simple computing elements controlled by its own software. The researchers compared the accuracy, speed and energy efficiency of SpiNNaker with that of NEST—a specialist supercomputer software currently in use for brain neuron-signaling research.

Steve Furber, co-author and Professor of Computer Engineering at the University of Manchester said, “The simulations run on NEST and SpiNNaker showed very similar results. This is the first time such a detailed simulation of the cortex has been run on SpiNNaker, or on any neuromorphic platform. SpiNNaker comprises 600 circuit boards incorporating over 500,000 small processors in total. The simulation described in this study used just six boards—1% of the total capability of the machine. The findings from our research will improve the software to reduce this to a single board.”

The researchers hope for increasingly large real-time simulations with these neuromorphic computing systems.

Cell-sized robots
MIT researchers have created what they believe may be the smallest robots yet that can sense their environment, store data, and even carry out computational tasks. The devices are about the size of a human egg cell, consisting of tiny electronic circuits made of 2D materials, piggybacking on minuscule particles called colloids.

Diagram illustrates the design of the tiny devices, which are designed to be able to float freely in liquid or air.

Source: MIT

Colloids — insoluble particles or molecules anywhere from a billionth to a millionth of a meter across — are so small they can stay suspended indefinitely in a liquid or even in air, and by coupling these tiny objects to complex circuitry, the researchers hope to lay the groundwork for devices that could be dispersed to carry out diagnostic journeys through anything from the human digestive system to oil and gas pipelines, or perhaps to waft through air to measure compounds inside a chemical processor or refinery.

Michael Strano, the Carbon C. Dubbs Professor of Chemical Engineering at MIT and senior author of the study, said “We wanted to figure out methods to graft complete, intact electronic circuits onto colloidal particles. Colloids can access environments and travel in ways that other materials can’t.”

He explained that dust particles can float indefinitely in the air because they are small enough that the random motions imparted by colliding air molecules are stronger than the pull of gravity. Similarly, colloids suspended in liquid will never settle out.

And while other groups have worked on the creation of similarly tiny robotic devices, their emphasis has been on developing ways to control movement, for example by replicating the tail-like flagellae that some microbial organisms use to propel themselves. But Strano suggested that may not be the most fruitful approach, since flagellae and other cellular movement systems are primarily used for local-scale positioning, rather than for significant movement. For most purposes, making such devices more functional is more important than making them mobile.

Interestingly, these tiny robots made by the MIT team are self-powered, requiring no external power source or even internal batteries. A simple photodiode provides the trickle of electricity that the tiny robots’ circuits require to power their computation and memory circuits, which is enough to let them sense information about their environment, store those data in their memory, and then later have the data read out after accomplishing their mission.

Such particles could potentially be used for diagnostic purposes in the human body, such as to pass through the digestive tract searching for signs of inflammation or other disease indicators, the researchers added.

However, most conventional microchips, such as silicon-based or CMOS, have a flat, rigid substrate and would not perform properly when attached to colloids that can experience complex mechanical stresses while travelling through the environment. In addition, all such chips are very energy-thirsty, the team said, which is why they tried the 2D electronic materials, including graphene and transition-metal dichalcogenides. They found these could be attached to colloid surfaces, remaining operational even after after being launched into air or water, and such thin-film electronics require only tiny amounts of energy.

Wearable cortisol sensor
The hormone cortisol rises and falls naturally throughout the day and can spike in response to stress but current methods for measuring cortisol levels require waiting several days for results from a lab, and by the time a person learns the results of a cortisol test – which may inform treatment for certain medical conditions – it is likely different from the current level of cortisol. Now, a group of Stanford University researchers led by materials scientist Alberto Salleo has created a stretchy patch that, applied directly to the skin, wicks up sweat and assesses how much cortisol a person is producing.

Drawing shows details of the layers contained in the cortisol biosensor developed by the Salleo lab and two close-up images of the holes in the bottom of the sensor that wick in sweat.
Source: Stanford University

This is particular interest in sweat sensing because it offers noninvasive and continuous monitoring of various biomarkers for a range of physiological conditions. This method offers a novel approach for the early detection of various diseases and evaluation of sports performance, the team said.

Clinical tests that measure cortisol provide an objective gauge of emotional or physical stress in research subjects and can help doctors tell if a patient’s adrenal or pituitary gland is working properly. If the prototype version of the wearable device becomes a reality, it could allow people with an imbalance to monitor their own levels at home, they said.

Further, a fast-working test like this could also reveal the emotional state of young – even non-verbal – children, who might not otherwise be able to communicate that they feel stress.

The researchers, hoping to take advantage of their generalizable design, are also figuring out what biomarker they may want to study next. Eventually, the goal would be to have a device that measures several biomarkers at once, which would give a clearer and more individualized picture of what is going on in a person’s body, they added.