Aurelia Butler

In a paper published today in Nature Photonics, researchers from the University of Oxford, along with collaborators from the Universities of Muenster, Heidelberg, and Exeter, report on their development of integrated photonic-electronic hardware capable of processing three-dimensional (3D) data, substantially boosting data processing parallelism for AI tasks.
Conventional computer chip processing efficiency doubles every 18 months, but the processing power required by modern AI tasks is currently doubling around every 3.5 months. This means that new computing paradigms are urgently needed to cope with the rising demand.
One approach is to use light instead of electronics — this allows multiple calculations to be carried out in parallel using different wavelengths to represent different sets of data. Indeed, in ground breaking work published in the journal Nature in 2021, many of the same authors demonstrated a form of integrated photonic processing chip that could carry out matrix vector multiplication (a crucial task for AI and machine learning applications) at speeds far outpacing the fastest electronic approaches. This work resulted in the birth of the photonic AI company, Salience Labs, a spin-out from the University of Oxford.
Now the team has gone further by adding an extra parallel dimension to the processing capability of their photonic matrix-vector multiplier chips. This “higher-dimensional” processing is enabled by exploiting multiple different radio frequencies to encode the data, propelling parallelism to a level far beyond that previously achieved.
As a test case the team applied their novel hardware to the task of assessing the risk of sudden death from electrocardiograms of heart disease patients. They were able to successfully analyse 100 electrocardiogram signals simultaneously, identifying the risk of sudden death with a 93.5% accuracy.
The researchers further estimated that even with a moderate scaling of 6 inputs × 6 outputs, this approach can outperform state-of-the-art electronic processors, potentially providing a 100-times enhancement in energy efficiency and compute density. The team anticipates further enhancement in computing parallelism in the future, by exploiting more degrees of freedom of light, such as polarization and mode multiplexing.
First author Dr Bowei Dong at the Department of Materials, University of Oxford said: ‘We previously assumed that using light instead of electronics could increase parallelism only by the use of different wavelengths — but then we realised that using radio frequencies to represent data opens up yet another dimension, enabling superfast parallel processing for emerging AI hardware.’
Professor Harish Bhaskaran, Department of Materials, University of Oxford and CO-founder of Salience Labs, who led the work said: ‘This is an exciting time to be doing research in AI hardware at the fundamental scale, and this work is one example of how what we assumed was a limit can be further surpassed.’ More

Researchers have invented an experimental wearable device that generates power from a user’s bending finger and can create and store memories, in a promising step towards health monitoring and other technologies.
The innovation features a single nanomaterial incorporated into a stretchable casing fitted to a person’s finger. The nanomaterial enabled the device to generate power with the user bending their finger.
The super-thin material also allows the device to perform memory tasks, as outlined below.
Multifunctional devices normally require several materials in layers, which involves the time-consuming challenge of stacking nanomaterials with high precision.
The team, led by RMIT University and the University of Melbourne in collaboration with other Australian and international institutions, made the proof-of-concept device with the rust of a low-temperature liquid metal called bismuth, which is safe and well suited for wearable applications.
Senior lead researcher Dr Ali Zavabeti said the invention could be developed to create medical wearables that monitor vital signs — incorporating the researchers’ recent work with a similar material that enabled gas sensing — and memorise personalised data.
“The innovation was used in our experiments to write, erase and re-write images in nanoscale, so it could feasibly be developed to one day encode bank notes, original art or authentication services,” said Zavabeti, an engineer from RMIT and the University of Melbourne. More

Robotic prosthetic ankles that are controlled by nerve impulses allow amputees to move more “naturally,” improving their stability, according to a new study from North Carolina State University and the University of North Carolina at Chapel Hill.
“This work focused on ‘postural control,’ which is surprisingly complicated,” says Helen Huang, corresponding author of the study and the Jackson Family Distinguished Professor in the Joint Department of Biomedical Engineering at NC State and UNC.
“Basically, when we are standing still, our bodies are constantly making adjustments in order to keep us stable. For example, if someone bumps into us when we are standing in line, our legs make a wide range of movements that we are not even necessarily aware of in order to keep us upright. We work with people who have lower limb amputations, and they tell us that achieving this sort of stability with prosthetic devices is a significant challenge. And this study demonstrates that robotic prosthetic ankles which are controlled using electromyographic (EMG) signals are exceptionally good at allowing users to achieve this natural stability.” EMG signals are the electrical signals recorded from an individual’s muscles.
The new study builds on previous work, which demonstrated that neural control of a powered prosthetic ankle can restore a range of abilities, including standing on challenging surfaces and squatting.
For this study, the researchers worked with five people who had amputations below the knee on one leg. Study participants were fitted with a prototype robotic prosthetic ankle that responds to EMG signals that are picked up by sensors on the leg.
“Basically, the sensors are placed over the muscles at the site of the amputation,” says Aaron Fleming, co-author of the study and recent Ph.D. graduate from NC State. “When a study participant thinks about moving the amputated limb, this sends electrical signals through the residual muscle in the lower limb. The sensors pick these signals up through the skin and translate those signals into commands for the prosthetic device.”
The researchers conducted general training for study participants using the prototype device, so that they were somewhat familiar with the technology. More

Taking inspiration from music streaming services, a team of engineers at the University of Michigan, Google and Georgia Tech has designed the simplest way for users to program their own exoskeleton assistance settings.
Of course, what’s simple for the users is more complex underneath, as a machine learning algorithm repeatedly offers pairs of assistance profiles that are most likely to be comfortable for the wearer. The user then selects one of these two, and the predictor offers another assistance profile that it believes might be better. This approach enables users to set the exoskeleton assistance based on their preferences using a very simple interface, conducive to implementing on a smartwatch or phone.
“It’s essentially like Pandora music,” said Elliott Rouse, U-M associate professor of robotics and mechanical engineering and corresponding author of the study in Science Robotics. “You give it feedback, a thumbs up or thumbs down, and it curates a radio station based on your feedback. This is a similar idea, but it’s with exoskeleton assistance settings. In both cases, we are creating a model of the user’s preferences and using this model to optimize the user’s experience.”
The team tested the approach with 14 participants, each wearing a pair of ankle exoskeletons as they walked at a steady pace of about 2.3 miles per hour. The volunteers could take as much time as they wanted between choices, although they were limited to 50 choices. Most participants were choosing the same assistance profile repeatedly by the 45th decision.
After 50 rounds, the experimental team began testing the users to see whether the final assistance profile was truly the best — pairing it against 10 randomly generated (but plausible) profiles. On average, participants chose the settings suggested by the algorithm about nine out of 10 times, which highlights the accuracy of the proposed approach.
“By using clever algorithms and a touch of AI, our system figures out what users want with easy yes-or-no questions,” said Ung Hee Lee, a recent U-M doctoral graduate from mechanical engineering and first author of the study, now at the robotics company Nuro. “I’m excited that this approach will make wearable robots comfortable and easy to use, bringing them closer to becoming a normal part of our day-to-day life.”
The control algorithm manages four exoskeleton settings: how much assistance to give (peak torque), how long to go between peaks (timing), and how the exoskeleton both ramps up and reduces the assistance on either side of each peak. This assistance approach is based on how our calf muscle adds force to propel us forward in each step. More

Ever since Antonie van Leeuwenhoek discovered the world of bacteria through a microscope in the late seventeenth century, humans have tried to look deeper into the world of the infinitesimally small.
There are, however, physical limits to how closely we can examine an object using traditional optical methods. This is known as the ‘diffraction limit’ and is determined by the fact that light manifests as a wave. It means a focused image can never be smaller than half the wavelength of light used to observe an object.
Attempts to break this limit with “super lenses” have all hit the hurdle of extreme visual losses, making the lenses opaque. Now physicists at the University of Sydney have shown a new pathway to achieve superlensing with minimal losses, breaking through the diffraction limit by a factor of nearly four times. The key to their success was to remove the super lens altogether.
The research is published today in Nature Communications.
The work should allow scientists to further improve super-resolution microscopy, the researchers say. It could advance imaging in fields as varied as cancer diagnostics, medical imaging, or archaeology and forensics.
Lead author of the research, Dr Alessandro Tuniz from the School of Physics and University of Sydney Nano Institute, said: “We have now developed a practical way to implement superlensing, without a super lens.
“To do this, we placed our light probe far away from the object and collected both high- and low-resolution information. By measuring further away, the probe doesn’t interfere with the high-resolution data, a feature of previous methods.”
Previous attempts have tried to make super lenses using novel materials. However, most materials absorb too much light to make the super lens useful. More

As the COVID-19 pandemic scattered and isolated people, researchers across Virginia Tech connected for a data-driven collaboration seeking improved drugs to fight the disease and potentially many other illnesses.
A multidisciplinary collaboration spanning several colleges at Virginia Tech resulted in a newly published study, “Data Driven Computational Design and Experimental Validation of Drugs for Accelerated Mitigation of Pandemic-like Scenarios,” in the Journal of Physical Chemistry Letters.
The study focuses on using computer algorithms to generate adaptations to molecules in compounds for existing and potential medications that can improve those molecules’ ability to bind to the main protease, a protein-based enzyme that breaks down complex proteins, in SARS-CoV-2, the virus that causes COVID-19.
This process allows exponentially more molecular adaptations to be considered than traditional trial-and-error methods of testing drugs one by one could allow. Candidate molecule adaptations can be identified among myriad possibilities, then narrowed to a few or one that can be created in a laboratory and tested for effectiveness.
“We present a novel transferable data-driven framework that can be used to accelerate the design of new small molecules and materials, with desired properties, by changing the combination of building blocks as well as decorating them with functional groups,” said Sanket A. Deshmukh, associate professor of chemical engineering in the College of Engineering. A “functional group” is a cluster of atoms that generally retains its characteristic properties, regardless of the other atoms in the molecule.
“Interestingly, the newly designed functionalized drug not only had a better half maximal effective concentration value than its parent drug, but also several of the proposed and used antivirals including Remdesivir,” Deshmukh said, referring to a measure of compound potency.
Moving through all the phases of the study would not have been possible without extensive cross-departmental collaboration. More

Artificial intelligence has exploded in popularity and is being harnessed by some scientists to predict which molecules could treat illnesses, or to quickly screen existing medicines for new applications. Researchers reporting in ACS Central Science have used one such deep learning algorithm, and found that dihydroartemisinin (DHA), an antimalarial drug and derivative of a traditional Chinese medicine, could treat osteoporosis as well. The team showed that in mice, DHA effectively reversed osteoporosis-related bone loss.
In healthy people, there is a balance between the osteoblasts that build new bone and osteoclasts that break it down. But when the “demolition crew” becomes overactive, it can result in bone loss and a disease called osteoporosis, which typically affects older adults. Current treatments for osteoporosis primarily focus on slowing the activity of osteoclasts. But osteoblasts — or more specifically, their precursors known as bone marrow mesenchymal stem cells (BMMSCs) — could be the basis for a different approach. During osteoporosis, these multipotent cells tend to turn into fat-creating cells instead, but they could be reprogrammed to help treat the disease. Previously, Zhengwei Xie and colleagues developed a deep learning algorithm that could predict how effectively certain small-molecule drugs reversed changes to gene expression associated with the disease. This time, joined by Yan Liu and Weiran Li, they wanted to use the algorithm to find a new treatment strategy for osteoporosis that focused on BMMSCs.
The team ran the program on a profile of differently expressed genes in newborn and adult mice. One of the top-ranked compounds identified was DHA, a derivative of artemisinin and a key component of malaria treatments. Administering DHA extract for six weeks to mice with induced osteoporosis significantly reduced bone loss in their femurs and nearly completely preserved bone structure. To improve delivery, the team designed a more robust system using injected, DHA-loaded nanoparticles. Bones of mice with osteoporosis that received the treatment were similar to those of the control group, and the treatment showed no evidence of toxicity. In further tests, the team determined that DHA interacted with BMMSCs to maintain their stemness and ultimately produce more osteoblasts. The researchers say that this work demonstrates that DHA is a promising therapeutic agent for osteoporosis.
The authors acknowledge funding from the National Natural Science Foundations of China, the Beijing International Science and Technology Cooperation, the Beijing Natural Science Foundation, Peking University Clinical Medicine Plus X — Young Scholars Project, the Ten-Thousand Talents Program, the Key R & D Plan of Ningxia Hui Autonomous Region, the Innovative Research Team of High-Level Local Universities in Shanghai, the Beijing Nova Program, the China National Postdoctoral Program for Innovative Talents, the China Postdoctoral Science Foundation, and the Peking University Medicine Sailing Program for Young Scholars’ Scientific & Technological Innovation. More

A collaborative research team, led by Professor Hyun-Kon Song in the School of Energy and Chemical Engineering at UNIST, Dr. Seo-Hyun Jung from Research Center for Advanced Specialty Chemicals at Korea Research Institute of Chemical Technology (KRICT), and Dr. Tae-Hee Kim from the Ulsan Advanced Energy Technology R&D Center at Korea Institute of Energy Research (KIER), has achieved a groundbreaking milestone in battery technology. Their remarkable achievement in developing a non-flammable gel polymer electrolyte (GPE) is set to revolutionize the safety of lithium-ion batteries (LIBs) by mitigating the risks of thermal runaway and fire incidents.
In the past, the potential flammability of LIBs has raised significant concerns, especially in electric vehicles, where fire hazards pose a serious threat to underground parking lots. Addressing this critical issue, the research team has successfully developed a groundbreaking non-flammable polymer semi-solid electrolyte, offering a promising solution to mitigate battery fires.
Conventionally, non-flammable electrolytes have heavily relied on the incorporation of flame retardant additives or solvents with exceptionally high boiling points. However, these methods often resulted in a considerable decrease in ion conductivity, compromising the overall performance of the electrolyte.
In their breakthrough research, the research team introduced a trace amount of polymer into the electrolyte, creating a semi-solid electrolyte. This novel approach dramatically increased the lithium ion conductivity by 33% compared to existing liquid electrolytes. Moreover, the pouch-type batteries incorporating this non-flammable semi-solid electrolyte exhibited a remarkable 110% improvement in life characteristics, effectively preventing unnecessary electrolyte reactions during the formation and operation of the solid-electrolyte interphase (SEI) layer.
The key advantage of this innovative electrolyte lies in its exceptional performance and non-combustibility. By suppressing radical chain reactions with fuel compounds during the combustion process, the polymer semi-solid electrolyte effectively inhibits the occurrence of battery fires. The research team demonstrated the excellence of the developed polymer by quantitatively analyzing its ability to stabilize and suppress radicals.
Jihong Jeong (School of Energy and Chemical Engineering, UNIST) emphasized, “The interaction between the polymerized material inside the battery and volatile solvents allows us to effectively suppress radical chain reactions. Through electrochemical quantification, this breakthrough will greatly contribute to understanding the mechanism of non-flammable electrolytes.”
Co-first author Mideum Kim, a master student in the School of Energy and Chemical Engineering at UNIST and the Korea Research Institute of Chemical Technology (KRICT), further confirmed the exceptional safety of the battery itself through various experiments. The team’s comprehensive approach included applying the non-flammable semi-solid electrolyte to pouch-type batteries, ensuring the evaluation of electrolyte non-combustibility extended to practical battery applications.
“The research team’s multidisciplinary composition, involving electrochemistry from UNIST, polymer synthesis from the KRICT Research Center for Advanced Specialty Chemicals, and battery safety testing by the Ulsan Advanced Energy Technology R&D Center at Korea Institute of Energy Research (KIER), has been instrumental in achieving this breakthrough,” stated Professor Song. “The use of non-flammable semi-solid electrolytes, which can be directly incorporated into existing battery assembly processes, will accelerate the future commercialization of safer batteries.”
The research study has applied for five patents in Korea and two overseas, further highlighting the significance of this achievement. Additionally, it has been selected as a supplementary cover for ACS Energy Letters, with publication online on October 13, 2023. This study has been made possible through the support of the National Research Foundation of Korea (NRF), the Ministry of Science and ICT (MSIT), the Korea Evaluation Institute of Industrial Technology (KEIT), the Korea Research Institute of Chemical Technology, and Samsung SDI Co., Ltd. More