Computers Math

Getting computers to “think” like humans is the holy grail of artificial intelligence, but human brains turn out to be tough acts to follow. The human brain is a master of applying previously learned knowledge to new situations and constantly refining what’s been learned. This ability to be adaptive has been hard to replicate in machines.
Now, Salk researchers have used a computational model of brain activity to simulate this process more accurately than ever before. The new model mimics how the brain’s prefrontal cortex uses a phenomenon known as “gating” to control the flow of information between different areas of neurons. It not only sheds light on the human brain, but could also inform the design of new artificial intelligence programs.
“If we can scale this model up to be used in more complex artificial intelligence systems, it might allow these systems to learn things faster or find new solutions to problems,” says Terrence Sejnowski, head of Salk’s Computational Neurobiology Laboratory and senior author of the new work, published on November 24, 2020, in Proceedings of the National Academy of Sciences.
The brains of humans and other mammals are known for their ability to quickly process stimuli — sights and sounds, for instance — and integrate any new information into things the brain already knows. This flexibility to apply knowledge to new situations and continuously learn over a lifetime has long been a goal of researchers designing machine learning programs or artificial brains. Historically, when a machine is taught to do one task, it’s difficult for the machine to learn how to adapt that knowledge to a similar task; instead each related process has to be taught individually.
In the current study, Sejnowski’s group designed a new computational modeling framework to replicate how neurons in the prefrontal cortex — the brain area responsible for decision-making and working memory — behave during a cognitive test known as the Wisconsin Card Sorting Test. In this task, participants have to sort cards by color, symbol or number — and constantly adapt their answers as the card-sorting rule changes. This test is used clinically to diagnose dementia and psychiatric illnesses but is also used by artificial intelligence researchers to gauge how well their computational models of the brain can replicate human behavior.
Previous models of the prefrontal cortex performed poorly on this task. The Sejnowski team’s framework, however, integrated how neurons control the flow of information throughout the entire prefrontal cortex via gating, delegating different pieces of information to different subregions of the network. Gating was thought to be important at a small scale — in controlling the flow of information within small clusters of similar cells — but the idea had never been integrated into models through the whole network.
The new network not only performed as reliably as humans on the Wisconsin Card Sorting Task, but also mimicked the mistakes seen in some patients. When sections of the model were removed, the system showed the same errors seen in patients with prefrontal cortex damage, such as that caused by trauma or dementia.
“I think one of the most exciting parts of this is that, using this sort of modeling framework, we’re getting a better idea of how the brain is organized,” says Ben Tsuda, a Salk graduate student and first author of the new paper. “That has implications for both machine learning and gaining a better understanding of some of these diseases that affect the prefrontal cortex.”
If researchers have a better understanding of how regions of the prefrontal cortex work together, he adds, that will help guide interventions to treat brain injury. It could suggest areas to target with deep brain stimulation, for instance.
“When you think about the ways in which the brain still surpasses state-of-the-art deep learning networks, one of those ways is versatility and generalizability across tasks with different rules,” says study coauthor Kay Tye, a professor in Salk’s Systems Neurobiology Laboratory and the Wylie Vale Chair. “In this new work, we show how gating of information can power our new and improved model of the prefrontal cortex.”
The team next wants to scale up the network to perform more complex tasks than the card-sorting test and determine whether the network-wide gating gives the artificial prefrontal cortex a better working memory in all situations. If the new approach works under broad learning scenarios, they suspect that it will lead to improved artificial intelligence systems that can be more adaptable to new situations.

Story Source:
Materials provided by Salk Institute. Note: Content may be edited for style and length. More

In the transition toward clean, renewable energy, there will still be a need for conventional power sources, like coal and natural gas, to ensure steady power to the grid. Researchers across the world are using unique materials and methods that will make those conventional power sources cleaner through carbon capture technology.
Creating accurate, detailed models is key to scaling up this important work. A recent paper led by the University of Pittsburgh Swanson School of Engineering examines and compares the various modeling approaches for hollow fiber membrane contactors (HFMCs), a type of carbon capture technology. The group analyzed over 150 cited studies of multiple modeling approaches to help researchers choose the technique best suited to their research.
“HFMCs are one of the leading technologies for post-combustion carbon capture, but we need modeling to better understand them,” said Katherine Hornbostel, assistant professor of mechanical engineering and materials science, whose lab led the analysis. “Our analysis can guide researchers whose work is integral to meeting our climate goals and help them scale up the technology for commercial use.”
A hollow fiber membrane contactor (HFMC) is a group of fibers in a bundle, with exhaust flowing on one side and a liquid solvent on the other to trap the carbon dioxide. The paper reviews state-of-the-art methods for modeling carbon capture HFMCs in one, two and three dimensions, comparing them in-depth and suggesting directions for future research.
“The ideal modeling technique varies depending on the project, but we found that 3D models are qualitatively different in the nature of information they can reveal,” said Joanna Rivero, graduate student working in the Hornbostel Lab and lead author. “Though cost limits their wide use, we identify 3D modeling and scale-up modeling as areas that will greatly accelerate the progress of this technology.”
Grigorios Panagakos, research engineer and teaching faculty in Carnegie Mellon University’s Department of Chemical Engineering, brought his expertise in analyzing the modeling of transport phenomena to the review paper, as well.

Story Source:
Materials provided by University of Pittsburgh. Note: Content may be edited for style and length. More

A team of researchers from the Technical University of Munich, the Walther-Meissner-Institute of the Bavarian Academy of Sciences and Humanities, and the Norwegian University of Science and Technology in Trondheim has discovered an exciting method for controlling spin carried by quantized spin wave excitations in antiferromagnetic insulators.
Elementary particles carry an intrinsic angular momentum known as their spin. For an electron, the spin can take only two particular values relative to a quantization axis, letting us denote them as spin-up and spin-down electrons. This intrinsic two-valuedness of the electron spin is at the core of many fascinating effects in physics.
In today’s information technology, the spin of an electron and the associated magnetic momentum are exploited in applications of information storage and readout of magnetic media, like hard disks and magnetic tapes.
Antiferromagnets: future stars in magnetic data storage?
Both, the storage media and the readout sensors utilize ferromagnetically ordered materials, where all magnetic moments align parallel. However, the moments may orient in a more complex way. In antiferromagnets, the “antagonist to a ferromagnet,” neighboring moments align in an anti-parallel fashion. While these systems look “non-magnetic” from outside, they have attracted broad attention as they promise robustness against external magnetic fields and faster control. Thus, they are considered as the new kids on the block for applications in magnetic storage and unconventional computing.
One important question in this context is, whether and how information can be transported and detected in antiferromagnets. Researchers at the Technical University of Munich, the Walther-Meissner-Institute and the Norwegian University of Science and Technology in Trondheim studied the antiferromagnetic insulator hematite in this respect.

advertisement

In this system, charge carriers are absent and therefore it is a particularly interesting testbed for the investigation of novel applications, where one aims at avoiding dissipation by a finite electrical resistance. The scientists discovered a new effect unique to the transport of antiferromagnetic excitations, which opens up new possibilities for information processing with antiferromagnets.
Unleashing the pseudospin in antiferromagnets
Dr Matthias Althammer, the lead researcher on the project describes the effect as follows: “In the antiferromagnetic phase, neighboring spins are aligned in an anti-parallel fashion. However, there are quantized excitations called magnons. Those carry information encoded in their spin and can propagate in the system. Due to the two antiparallel-coupled spin species in the antiferromagnet the excitation is of a complex nature, however, its properties can be cast in an effective spin, a pseudospin. We could experimentally demonstrate that we can manipulate this pseudospin, and its propagation with a magnetic field.”
Dr Akashdeep Kamra, the lead theoretician from NTNU in Trondheim adds that “this mapping of the excitations of an antiferromagnet onto a pseudospin enables an understanding and a powerful approach which has been the crucial foundation for treating transport phenomena in electronic systems. In our case, this enables us to describe the dynamics of the system in a much easier manner, but still maintain a full quantitative description of the system. Most importantly, the experiments provide a proof-of-concept for the pseudospin, a concept which is closely related to fundamental quantum mechanics.”
Unlocking the full potential of antiferromagnetic magnons
This first experimental demonstration of magnon pseudospin dynamics in an antiferromagnetic insulator not only confirms the theoretical conjectures on magnon transport in antiferromagnets, but also provides an experimental platform for expanding towards rich electronics inspired phenomena.
“We may be able to realize fascinating new stuff such as the magnon analogue of a topological insulator in antiferromagnetic materials” points out Rudolf Gross, director of the Walther-Meissner-Institute, Professor for Technical Physics (E23) at the Technical University of Munich and co-speaker for the cluster of excellence Munich Center for Quantum Science and Technology (MCQST). “Our work provides an exciting perspective for quantum applications based on magnons in antiferromagnets”
The research was funded by the Deutsche Forschungsgemeinschaft (DFG) via the cluster of excellence Munich Center for Quantum Science and Technology (MCQST) and by the Research Council of Norway. More

Researchers at the University of Tsukuba have created a new carbon-based electrical device, π-ion gel transistors (PIGTs), by using an ionic gel made of a conductive polymer. This work may lead to cheaper and more reliable flexible printable electronics.
Organic conductors, which are carbon-based polymers that can carry electrical currents, have the potential to radically change the way electronic devices are manufactured. These conductors have properties that can be tuned via chemical modification and may be easily printed as circuits. Compared with current silicon solar panels and transistors, systems based on organic conductors could be flexible and easier to install. However, their electrical conductivity can be drastically reduced if the conjugated polymer chains become disordered because of incorrect processing, which greatly limits their ability to compete with existing technologies.
Now, a team of researchers led by the University of Tsukuba have formulated a novel method for preserving the electrical properties of organic conductors by forming an “ion gel.” In this case, the solvent around the poly(para-phenyleneethynylene) (PPE) chains was replaced with an ionic liquid, which then turned into a gel. Using confocal fluorescent microscopy and scanning electron microscopy, the researchers were able to verify the morphology of the organic conductor.
“We showed that the internal structure of our π-ion gel is a nanofiber network of PPE, which is very good at reliably conducting electricity” says author Professor Yohei Yamamoto.
In addition to acting as wires for delocalized electrons, the polymer chains direct the flow of mobile ions, which can help move charge-carriers to the carbon rings. This allows current to flow through the entire volume of the device. The resulting transistor can switch on and off in response to voltage changes in less than 20 microseconds — which is faster than any previous device of this type.
“We plan to use this advance in supramolecular chemistry and organic electronics to design a whole arrange of flexible electronic devices,” explains Professor Yamamoto. The fast response time and high conductivity open the way for flexible sensors that enjoy the ease of fabrication associated with organic conductors, without sacrificing speed or performance.

Story Source:
Materials provided by University of Tsukuba. Note: Content may be edited for style and length. More

A model based solely on the past 40 years of weather events uses 7,000 times less computer power than today’s weather forecasting tools. An A.I.-powered model could someday provide more accurate forecasts for rain, snow and other weather events. More

New research offers clues to what goes on inside the minds of machines as they learn to see. Instead of attempting to account for a neural network’s decision-making on a post hoc basis, their method shows how the network learns along the way, by revealing how much the network calls to mind different concepts to help decipher what it sees as the image travels through successive layers. More

Neuroscientists have found reading computer code does not rely on the regions of the brain involved in language processing. Instead, it activates the ‘multiple demand network,’ which is also recruited for complex cognitive tasks such as solving math problems or crossword puzzles. More

Scientists often look to nature for cues when designing robots – some robots mimic human hands while others simulate the actions of octopus arms or inchworms. Now, researchers have designed a new soft robotic gripper that draws inspiration from an unusual source: pole beans. More