More stories

  • in

    Digitally removing clouds from aerial images using machine learning

    Scientists from the Division of Sustainable Energy and Environmental Engineering at Osaka University used an established machine learning technique called generative adversarial networks to digitally remove clouds from aerial images. By using the resulting data as textures for 3D models, more accurate datasets of building image masks can be automatically generated. When setting two artificial intelligence networks against each other, the team was able to improve the data quality without the need for previously labeled images. This work may help automate computer vision jobs critical to civil engineering.
    Machine learning is a powerful method for accomplishing artificial intelligence tasks, such as filling in missing information. One popular application is repairing images that are obscured, for example, when aerial images of buildings are blocked by clouds. While this can be done by hand, it is very time consuming, and even the machine learning algorithms that are currently available require many training images in order to work. Thus, improving the representation of buildings in virtual 3D models using aerial photographs requires additional steps.
    Now, researchers at Osaka University have improved the accuracy of automatically generated datasets by applying the existing machine learning method called generative adversarial networks (GANs). The idea of GANs is to pit two different algorithms against each other. One is the “generative network,” that proposes reconstructed images without clouds. Competing against it is the “discriminative network,” that uses a convolutional neural network to attempt to tell the difference between the digitally repaired pictures and actual images without clouds. Over time, both networks get increasingly better at their respective jobs, leading to highly realistic images with the clouds digitally erased. “By training the generative network to ‘fool’ the discriminative network into thinking an image is real, we obtain reconstructed images that are more self-consistent,” first author Kazunosuke Ikeno explains.
    The team used 3D virtual models with photographs from an open-source dataset as input. This allowed for the automatic generation of digital “masks” that overlaid reconstructed buildings over the clouds. “This method makes it possible to detect buildings in areas without labeled training data,” senior author Tomohiro Fukuda says. The trained model could detect buildings with an “intersection over union” value of 0.651, which measures how accurately the reconstructed area corresponds to the actual area. This method can be extended to improving the quality of other datasets in which some areas are obscured, such as medical images.
    Story Source:
    Materials provided by Osaka University. Note: Content may be edited for style and length. More

  • in

    Can a piece of sticky tape stop computer hackers in their tracks?

    Researchers from the University of Technology Sydney (UTS) and TMOS, an Australian Research Council Centre of Excellence, have taken the fight to online hackers with a giant leap towards realizing affordable, accessible quantum communications, a technology that would effectively prevent the decryption of online activity. Everything from private social media messaging to banking could become more secure due to new technology created with a humble piece of adhesive tape.
    Quantum communication is still in its early development and is currently feasible only in very limited fields due to the costs associated with fabricating the required devices. The TMOS researches have developed new technology that integrates quantum sources and waveguides on chip in a manner that is both affordable and scalable, paving the way for future everyday use.
    The development of fully functional quantum communication technologies has previously been hampered by the lack of reliable quantum light sources that can encode and transmit the information.
    In a paper published today in ACS Photonics, the team describes a new platform to generate these quantum emitters based on hexagonal boron nitride, also known as white graphene. Where current quantum emitters are created using complex methods in expensive clean rooms, these new quantum emitters can be created using $20 worth of white graphene pressed on to a piece of adhesive tape.
    These 2D materials can be pressed onto a sticky surface such as the adhesive tape and exfoliated, which is essentially peeling off the top layer to create a flex. Multiple layers of this flex can then be assembled in a Lego-like style, offering a new bottom up approach as a substitute for 3D systems.
    TMOS Chief Investigator Igor Aharonovich said: “2D materials, like hexagonal boron nitride, are emerging materials for integrated quantum photonics, and are poised to impact the way we design and engineer future optical components for secured communication.”
    In addition to this evolution in photon sources, the team has developed a high efficiency on-chip waveguide, a vital component for on-chip optical processing.
    Lead author Chi Li said: “Low signal levels have been a significant barrier preventing quantum communications from evolving into practical, workable models. We hope that with this new development, quantum comms will become an everyday technology that improves people’s lives in new and exciting ways.”
    Story Source:
    Materials provided by University of Technology Sydney. Note: Content may be edited for style and length. More

  • in

    Promising candidates revealed for next-generation LED-based data communications

    A new paper from the University of Surrey and the University of Cambridge has detailed how two relatively unexplored semiconducting materials can satisfy the telecommunication industry’s hunger for enormous amounts of data at ever-greater speeds.
    Light-emitting diode (LED)-based communications techniques allow computing devices, including mobile phones, to communicate with one another by using infrared light. However, LED techniques are underused because in its current state LED transmits data at far slower speeds than other wireless technologies such as light-fidelity (Li-Fi).
    In a paper published by Nature Electronics, the researchers from Surrey and Cambridge, along with partners from the University of Electronic Science and Technology of China, examine how organic semiconductors, colloidal quantum dots (CQDs) and metal halide perovskites (perovskites), can be used in LED-based optical communications systems.
    The research team explored efforts to improve the performance and efficiency of these LEDs, and they considered their potential applications in on-chip interconnects and Li-Fi.
    Dr Aobo Ren, the co-first author and visiting postdoctoral researcher at the University of Surrey, said:
    “There’s excitement surrounding CQDs and perovskites because they offer great promise for low-power, cost-effective and scalable communications modules. More

  • in

    Mobile alert app for missing people with dementia

    Researchers are working with community leaders to develop a mobile alert app to help locate missing people with dementia.
    Noelannah Neubauer, a postdoctoral fellow at the University of Waterloo’s School of Public Health Sciences, said “Community ASAP” is aimed at addressing a gap in available tools when it comes to alerts for missing older adults and people living with dementia in Canada.
    “We have Amber Alerts for missing children, but nothing for this population other than police and civilians circulating information via social media such as Twitter and Facebook,” said Neubauer, who is the first author of a study that tested the efficacy and useability of Community ASAP.
    The U.S. already uses a system called Silver Alert, and there have been efforts in some provinces, such as British Columbia, to create a citizen-led alert system. The issue with piggybacking onto the Amber Alert system is that too many people go missing every day, according to Neubauer. Almost 750,000 Canadians live with dementia, and 60 per cent of them wander at least once, and some repeatedly.
    “The sheer number of missing people from this population would mean that alerts would go off multiple times a day in certain jurisdictions, running the risk of significant alert fatigue,” Neubauer said. “Community ASAP gets around this by having people sign up to receive the alert on Android and iOS operating systems and choosing the radius from where the missing person was last seen to their current location. Most missing cases take place one kilometre from the place they were last seen.”
    “A key concern is that if someone gets lost and is not found within 24 hours, they have a 50 per cent chance of experiencing serious injury or death,” said Lili Liu, principal investigator, and dean of the Faculty of Health at Waterloo. “We proposed recommendations for community alert systems specific to Canada, such as Community ASAP, at an online national forum on community alert systems for missing older adults last fall.”
    For the study, researchers engaged people living with dementia, their care partners, police services, search and rescue organizations and health and social service providers in Ontario, Alberta and British Columbia to develop the alert system that engages community citizens, as volunteers, to look out for people with dementia reported missing.
    They went through three iterations of the app and consulted with these stakeholder groups along the way to test its accuracy and useability, walking through scenarios to simulate the events that transpire during a missing person event. In these scenarios, participants assumed the key roles in the Community ASAP system, including the missing person with dementia, care partner, coordinator, and volunteers.
    The idea for this app came from Ron Beleno, an entrepreneur with experience caring for his father, who lived with dementia. Beleno is turning Community ASAP into a start-up company, and Liu’s research team continues to work with governments and organizations to coordinate a system that works across the country.
    The study, “Mobile alert app to engage community volunteers to help locate missing persons with dementia,” co-authored by Noelannah Neubauer, Christine Daum, Antonio Miguel-Cruz and Lili Liu, all affiliated with the University of Waterloo, was recently published in Plos One. More

  • in

    Artificial intelligence to help predict Arctic sea ice loss

    A new AI (artificial intelligence) tool is set to enable scientists to more accurately forecast Arctic sea ice conditions months into the future. The improved predictions could underpin new early-warning systems that protect Arctic wildlife and coastal communities from the impacts of sea ice loss.
    Published this week (Thursday 26 August) in the journal Nature Communications, an international team of researchers led by British Antarctic Survey (BAS) and The Alan Turing Institute describe how the AI system, IceNet, addresses the challenge of producing accurate Arctic sea ice forecasts for the season ahead — something that has eluded scientists for decades.
    Sea ice, a vast layer of frozen sea water that appears at the North and South poles, is notoriously difficult to forecast because of its complex relationship with the atmosphere above and ocean below. The sensitivity of sea ice to increasing temperatures has caused the summer Arctic sea ice area to halve over the past four decades, equivalent to the loss of an area around 25 times the size of Great Britain. These accelerating changes have dramatic consequences for our climate, for Arctic ecosystems, and Indigenous and local communities whose livelihoods are tied to the seasonal sea ice cycle.
    IceNet, the AI predictive tool, is almost 95% accurate in predicting whether sea ice will be present two months ahead — better than the leading physics-based model.
    Lead author Tom Andersson, Data Scientist at the BAS AI Lab and funded by The Alan Turing Institute, explains: “The Arctic is a region on the frontline of climate change and has seen substantial warming over the last 40 years. IceNet has the potential to fill an urgent gap in forecasting sea ice for Arctic sustainability efforts and runs thousands of times faster than traditional methods.”
    Dr Scott Hosking, Principal Investigator, Co-leader of the BAS AI Lab and Senior Research Fellow at The Alan Turing Institute, says: “I’m excited to see how AI is making us rethink how we undertake environmental research. Our new sea ice forecasting framework fuses data from satellite sensors with the output of climate models in ways traditional systems simply couldn’t achieve.”
    Unlike conventional forecasting systems that attempt to model the laws of physics directly, the authors designed IceNet based on a concept called deep learning. Through this approach, the model ‘learns’ how sea ice changes from thousands of years of climate simulation data, along with decades of observational data to predict the extent of Arctic sea ice months into the future.
    Tom Andersson concludes: “Now we’ve demonstrated that AI can accurately forecast sea ice, our next goal is to develop a daily version of the model and have it running publicly in real-time, just like weather forecasts. This could operate as an early warning system for risks associated with rapid sea ice loss.”
    Story Source:
    Materials provided by British Antarctic Survey. Note: Content may be edited for style and length. More

  • in

    MOGONET provides more holistic view of biological processes underlying disease

    Genomics, proteomics, metabolomics, transcriptomics — rapid advances in high-throughput biomedical technologies has enabled the collection of data with unprecedented detail from the growing number of omics. But, how best to take advantage of the interactions and complementary information in omics data?
    To fully utilize the advances in omics technologies to achieve a more comprehensive understanding of the biological processes underlying human diseases, researchers from Regenstrief Institute and Indiana, Purdue and Tulane Universities have developed and tested MOGONET, a novel multi-omics data analysis algorithm and computational methodology. Integrating data from various omics provides a more holistic view of biological processes underlying human diseases. The creators have made MOGONET open source, free and accessible to all researchers.
    In a study published in Nature Communications, the scientists demonstrated that MOGONET, short for Multi-Omics Graph cOnvolutional NETworks, outperforms existing supervised multi-omics integrative analysis approaches of different biomedical classification applications using mRNA expression data, DNA methylation data, and microRNA expression data.
    They also determined that MOGONET can identify important omics signatures and biomarkers from different omics data types.
    “With MOGONET, our new AI [artificial intelligence] tool, we employ machine learning based on a neural network, to capture complex biological process relationships. We have made the understanding of omics more comprehensive and also are learning more about disease subtypes that biomarkers help us differentiate,” said Regenstrief Institute Research Scientist Kun Huang, PhD, who led the study. “The ultimate goal is to improve disease prognosis and enhance disease-outcome predictions.” A bioinformatician, he credits the diversity of the MOGONET research group, which included computer scientists as well as data scientists and bioinformaticians, with their varying perspectives, as instrumental in its development and success. He serves as director of data sciences and informatics for the Indiana University Precision Health Initiative.
    The researchers tested MOGONET on datasets related to o Alzheimer’s disease, gliomas, kidney cancer and breast invasive carcinoma as well as on healthy patient datasets. They determined MOGONET handily outperformed existing supervised multi-omics integration methods.
    “Learning and integrating intuitive recognition, MOGONET could generate new biomarker disease candidates,”said study co-author Regenstrief Institute Affiliated Scientist Jie Zhang, PhD, a bioinformatician. “MOGONET also could predict new cancer subtypes, tumor grade and disease progression. It can identify normal brain activity versus Alzheimer’s disease.”
    Drs. Huang and Zhang plan to expand this work beyond omics to include imaging data, noting the abundance of brain images for AD and cancer-related pathology images which can teach MOGONET to recognize even cases it had not previously encountered. Both scientists note that following rigorous clinical studies, MOGONET could support improved patient care in many areas.
    In addition to Drs. Huang and Zhang, authors of “MOGONET integrates multi-omics data using graph convolutional networks allowing patient classification and biomarker identification” are Tongxin Wang, PhD, and Haixu Tang, PhD, of Indiana University, Wei Shao, PhD, of IU School of Medicine; Zhi Huang of IU School of Medicine and Purdue University; and Zhengming Ding, PhD of Tulane University. Dr. Wang worked in Dr. Huang’s laboratory. Dr. Ding, formerly of Indiana University, is an expert in the field of machine learning.
    The development and testing of MOGONET was supported by National Institutes of Health grants R01EB025018 and U54AG065181 and the Indiana University Precision Health Initiative.
    Story Source:
    Materials provided by Regenstrief Institute. Note: Content may be edited for style and length. More

  • in

    Physical activity in children can be improved through ‘exergames’

    Physical activity among young people can be improved by well-designed and delivered online interventions such as ‘exergames’ and smartphone apps, new research shows.
    According to a review study carried out at the University of Birmingham, children and young people reacted positively in PE lessons to the use of exergames, which deliver physical activity lessons via games or personalised activities. Changes included increases in physical activity levels, but also improved emotions, attitudes and motivations towards physical activity.
    The study, published in Physical Education and Sport Pedagogy is one of the first to examine not only the impact of online interventions on physical behaviours in non-clinical groups of young people but the effects of digital mediums on physical activity knowledge, social development and improving mental health.
    The evidence can be used to inform guidance for health and education organisations on how they can design online interventions to reach and engage young people in physical activity.
    The authors analysed 26 studies of online interventions for physical activity. They found three main mechanisms at work: gamification, in which participants progress through different levels of achievement; personalisation, in which participants received tailored feedback and rewards based on progress; and information, in which participants received educational material or guidance to encourage behavioural change.
    Most of the interventions were focused on gamification or personalisation and the researchers found the majority of studies (70%) reported an increase and/or improvement in outcomes related to physical activity for children and young people who participated in online interventions. Primary school age pupils in particular who participated during PE lessons benefited.
    Lead author Dr Victoria Goodyear, in the University of Birmingham’s School of Sport, Exercise and Rehabilitation Science, said: “We find convincing evidence that PE teachers can use online learning to boost attitudes and participation in physical activity among young people, particularly at primary school age. There’s a real opportunity here for the PE profession to lead the way in designing meaningful and effective online exercise opportunities, as well as an opportunity to embed positive approaches to exercise and online games and apps at an early stage.”
    Story Source:
    Materials provided by University of Birmingham. Note: Content may be edited for style and length. More

  • in

    Robot mimics the powerful punch of the mantis shrimp

    Mantis shrimp pack the strongest punch of any creature in the animal kingdom. Their club-like appendages accelerate faster than a bullet out of a gun and just one strike can knock the arm off a crab or break through a snail shell. These small but mighty crustaceans have been known to take on octopus and win.
    How mantis shrimp produce these deadly, ultra-fast movements has long fascinated biologists. Recent advancements in high-speed imaging make it possible to see and measure these strikes but some of the mechanics have not been well understood.
    Now, an interdisciplinary team of roboticists, engineers and biologists have modeled the mechanics of the mantis shrimp’s punch and built a robot that mimics the movement. The research sheds light on the biology of these pugnacious crustaceans and paves the way for small but mighty robotic devices.
    The research is published in the Proceedings of the National Academy of Sciences.
    “We are fascinated by so many remarkable behaviors we see in nature, in particular when these behaviors meet or exceed what can be achieved by human-made devices,” said Robert Wood, the Harry Lewis and Marlyn McGrath Professor of Engineering and Applied Sciences at the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) and senior author of the paper. “The speed and force of mantis shrimp strikes, for example, are a consequence of a complex underlying mechanism. By constructing a robotic model of a mantis shrimp striking appendage, we are able to study these mechanisms in unprecedented detail.”
    Many small organisms — including frogs, chameleons, even some kinds of plants — produce ultra-fast movements by storing elastic energy and rapidly releasing it through a latching mechanism, like a mouse trap. In mantis shrimp, two small structures embedded in the tendons of the muscles called sclerites act as the appendage’s latch. In a typical spring-loaded mechanism, once the physical latch is removed, the spring would immediately release the stored energy. More