Goto

Collaborating Authors

South Holland


Moon exploration rovers from Lunar Zebro. What if you would use this?

Robohub

In this series of articles, we take robot innovations from their test-lab and bring them to a randomly selected workplace in the outside world. We discovered that Lunar Zebro is not only good for risky endeavours on extraterrestrial terrains, but that these sturdy self-organising little rovers can also simplify the life of hardworking interior decorators. 'Shoot for the moon' is taken quite literally by this team of students and professors from TU Delft. "We want to make the exploration of the moon available for a wider audience", says Pieter van Santvliet, partnerships coordinator at Lunar Zebro. "So we build the world's smallest and lightest rover."


Robots flying a kite to generate electricity. What if you would use this?

Robohub

Since 2007, two professors at the TU Delft have been researching ways to harvest energy from the wind using a kite. The robotic kite looks set to make its debut in the energy sector, but often inventions are used in unexpected ways. In this series of articles, we take robot innovations from their test-lab and bring them to a randomly selected workplace in the outside world. From kindergarten teacher Fransien, we learn that big kites could also be child's play, quite literally. A robot wheels in the kite and then slowly releases it, painting 8-shaped loops on the sky.


Study finds that even the best speech recognition systems exhibit bias

#artificialintelligence

Even state-of-the-art automatic speech recognition (ASR) algorithms struggle to recognize the accents of people from certain regions of the world. That's the top-line finding of a new study published by researchers at the University of Amsterdam, the Netherlands Cancer Institute, and the Delft University of Technology, which found that an ASR system for the Dutch language recognized speakers of specific age groups, genders, and countries of origin better than others. Speech recognition has come a long way since IBM's Shoebox machine and Worlds of Wonder's Julie doll. But despite progress made possible by AI, voice recognition systems today are at best imperfect -- and at worst discriminatory. In a study commissioned by the Washington Post, popular smart speakers made by Google and Amazon were 30% less likely to understand non-American accents than those of native-born users.


Even the Best Speech Recognition Systems Exhibit Bias, Study Finds - Slashdot

#artificialintelligence

An anonymous reader quotes a report from VentureBeat: Even state-of-the-art automatic speech recognition (ASR) algorithms struggle to recognize the accents of people from certain regions of the world. That's the top-line finding of a new study published by researchers at the University of Amsterdam, the Netherlands Cancer Institute, and the Delft University of Technology, which found that an ASR system for the Dutch language recognized speakers of specific age groups, genders, and countries of origin better than others. The coauthors of this latest research set out to investigate how well an ASR system for Dutch recognizes speech from different groups of speakers. In a series of experiments, they observed whether the ASR system could contend with diversity in speech along the dimensions of gender, age, and accent. The researchers began by having an ASR system ingest sample data from CGN, an annotated corpus used to train AI language models to recognize the Dutch language.


Machine learning helps in the search for new antibiotics

AIHub

With the search for new antibiotics becoming increasingly urgent, artificial intelligence offers valuable help. Smart software developed by Leiden PhD candidate Alexander Kloosterman searched genomes of bacteria and found clusters of DNA that code for proteins that have an antibiotic effect. "This new search method is enormously promising." The discovery was published in PLoS Biology. Professor of Molecular Biotechnology Gilles van Wezel from the Leiden Institute of Biology (IBL) initiated the research together with visiting professor Marnix Medema.


PhD Position In Artificial Intelligence At Delft University Of Technology 2021-2022 - AI Summary

#artificialintelligence

The importance of considering distributive justice in climate policy motivates research in AI-based decision support to search for balanced alternatives across multiple sectors, regions, and generations and counteract existing asymmetries in policy design. This PhD position is one of the four PhD positions in the Hippo Lab (Hyper-heuristics for interpretable public policy analysis), which is part of the TU Delft Artificial Intelligence initiative to channel expertise in AI foundations to tackle societal and scientific challenges. With its excellent education and research at the intersection of technology, society and policy, the Faculty of TPM contributes to solving complex technical-social issues, such as energy transition, mobility, digitalisation, water management and (cyber) security. Stay updated on last news about Artificial Intelligence. Check your inbox or spam folder to confirm your subscription.


Demo Day: the most exciting day of the year (Jan 28th, 3pm CET)

Robohub

Multidisciplinary teams of minor students have designed and built a functional robotic prototype for a project customer with a challenge to improve the lives of working people. Streaming live from RoboHouse on the TU Delft Campus, you can join a virtual celebration of ideas en technology with some of our communities finest talent. On Thursday 28 January 3pm (CET) you can visit the streaming website that we have created for the occasion. There you will find live video streams for a programme with a project overview and demonstrations by each student team. This all happens between 3pm and 4:30pm.


AI can tell what song you are listening to from your brainwaves

New Scientist

ARTIFICIAL intelligence has learned to identify the songs someone is listening to from their brain readings. Derek Lomas at Delft University of Technology in the Netherlands and his colleagues asked 20 people to listen to 12 songs through headphones. The volunteers did this blindfolded and in a dimly lit room to minimise the effect of their other senses on the results. Each person's brainwaves were recorded using an electroencephalography (EEG) cap that detects electrical activity.


How to keep drones flying when a motor fails

Robohub

Robotics researchers at the University of Zurich show how onboard cameras can be used to keep damaged quadcopters in the air and flying stably – even without GPS. As anxious passengers are often reassured, commercial aircrafts can easily continue to fly even if one of the engines stops working. But for drones with four propellers – also known as quadcopters – the failure of one motor is a bigger problem. With only three rotors working, the drone loses stability and inevitably crashes unless an emergency control strategy sets in. Researchers at the University of Zurich and the Delft University of Technology have now found a solution to this problem: They show that information from onboard cameras can be used to stabilize the drone and keep it flying autonomously after one rotor suddenly gives out.


Open access takes flight

Science

As a new mandate takes effect, researchers and institutions grapple with the trade-offs of making scientific publications free for all In 2018, a group of mostly European funders sent shock waves through the world of scientific publishing by proposing an unprecedented rule: The scientists they funded would be required to make journal articles developed with their support immediately free to read when published. The new requirement, which takes effect starting this month, seeks to upend decades of tradition in scientific publishing, whereby scientists publish their research in journals for free and publishers make money by charging universities and other institutions for subscriptions. Advocates of the new scheme, called Plan S (the “S” stands for the intended “shock” to the status quo), hope to destroy subscription paywalls and speed scientific progress by allowing findings to be shared more freely. It's part of a larger shift in scientific communication that began more than 20 years ago and has recently picked up steam. Scientists have several ways to comply with Plan S, including by paying publishers a fee to make an article freely available on a journal website, or depositing the article in a free public repository where anyone can download it. The mandate is the first by an international coalition of funders, which now includes 17 agencies and six foundations, including the Wellcome Trust and Howard Hughes Medical Institute, two of the world's largest funders of biomedical research. The group, which calls itself Coalition S, has fallen short of its initial aspiration to catalyze a truly international movement, however. Officials in three top producers of scientific papers—China, India, and the United States—have expressed general support for open access, but have not signed on to Plan S. Its mandate for immediate open access will apply to authors who produced only about 6% of the world's papers in 2017, according to an estimate by the Clarivate analytics firm, publisher of the Web of Science database. Still, there's reason to think Coalition S will make an outsize impact, says Johan Rooryck, Coalition S's executive director and a linguist at Leiden University. In 2017, 35% of papers published in Nature and 31% of those in Science cited at least one coalition member as a funding source. “The people who get [Coalition S] funding are very prominent scientists who put out very visible papers,” Rooryck says. “We punch above our weight.” In a dramatic sign of that influence, the Nature and Cell Press families of journals—stables of high-profile publications—announced in recent weeks that they would allow authors to publish papers outside their paywall, for hefty fees. Other recent developments point to growing support for open access. In 2017, for the first time, the majority of new papers across all scholarly disciplines, most of them in the sciences, were published open access, according to the Curtin Open Knowledge Initiative. More recently, most major publishers removed paywalls from articles about COVID-19 last year in an attempt to speed development of vaccines and treatments. Despite these and other signs of momentum, some publishing specialists say Plan S and other open-access measures could be financially stressful and ultimately unsustainable for publishers and the research institutions and authors who foot the bill. As debate continues about just how far and fast the movement will go, Science offers this guide for authors readying to plunge in. Authors who make their work open access may reap benefits, but their magnitude depends partly on what you measure. One yardstick is a paper's impact. Some studies have reported up to triple the number of citations for open-access articles on average compared with paywalled ones. But authors may be likely to publish their best work open access, which might bring it more citations. A recent analysis that used statistical methods to control for this tendency found a far more modest citation advantage for open access—8%—and only for a minority of “superstar” papers. Mark McCabe of SKEMA Business School and Christopher Snyder of Dartmouth College studied how citations to articles changed when their journal volumes moved from behind paywalls to entirely open access, and compared them with citations for articles that remained paywalled. For each article in their sample of more than 200,000 papers in ecology and other fields, the researchers accounted for other characteristics that affect citations, such as a paper's age: Newly published papers usually receive a burst of citations at first but fewer later. The modest citation advantage from open access accrued only to high-quality papers, defined as having already garnered 11 or more citations during a 2-year period before the paper became open access, McCabe and Snyder reported in November 2020. Other studies have found that open-access articles have a larger reach by other measures, including the number of downloads and online views. They also have an edge in Altmetric scores, a composite of an article's mentions on social media and in news stories and policy documents. These nonscholarly mentions buttress reports that open access enables a broader audience, beyond the core scientific community, to read research findings. In November 2020, Springer Nature and partners released findings from a survey of 6000 visitors to its websites. They reported that an “astonishing” 28% were general users, including patients, teachers, and lawyers. Another 15% worked in industry or medical jobs that required them to read but not publish research. Even for faculty members who can read subscription-based journals through their institution's libraries, open access could allow quicker access to articles in journals to which the institution doesn't subscribe. Some 57% of academics surveyed said they “almost always” or “frequently” had trouble accessing the full content of Springer Nature's articles. Open access comes in different varieties, or colors, each with its own costs and benefits. In what's called gold open access, articles carry a license making them freely available on publication. Typically the publisher charges a fee to offset lost subscription revenue and cover the cost of publishing. In recent years, the median paid, after discounts, was about $2600, according to a 2020 study by Nina Schönfelder of Bielefeld University. More selective journals, such as The Lancet Global Health , have charged up to $5000. The Nature Research family of journals has set its top open-access fee at €9500 (about $11,600), and Cell Press will charge $9900 for its flagship, Cell . Some journals are entirely gold open access; other, “hybrid” journals offer authors a choice between free publication behind a paywall or open access for a fee. A growing number of universities and research institutions, especially in Europe, are striking deals in which they pay a publisher a single fee that covers open-access publishing by their authors and also lets people on their campuses read content that remains behind paywalls. The largest such agreement was reached in 2019 between Springer Nature and 700 German research institutions and libraries. Since the first such deal in 2015, the number grew to 137 in 2020, according to the ESAC Transformative Agreement Registry. However, the deals last year covered publication fees for only 3% of papers produced globally. A variant called green open access allows authors to avoid publication fees. In this arrangement, authors publish in journals—even ones that use paywalls instead of charging authors—but also make their article freely available in an online repository. U.S. policy already requires the final, published versions of papers developed with federal funding to be deposited within 12 months in a repository such as the National Institutes of Health's PubMed Central, and many publishers do this automatically. Other authors can use online tools to find repositories. The Directory of Open Access Repositories lists more than 5500 of them. Publishers typically impose a 6- or 12-month embargo before authors can deposit the final, peer-reviewed version of a paywalled article, but this runs afoul of the Plan S requirement for immediate open access. (The embargo policies of thousands of journals globally are listed in a database called Sherpa/Romeo.) As a compromise, many publishers including the Science family of journals allow authors to immediately post a nearly final, peer-reviewed version of a paper in an institutional repository. Plan S accepts this form of green open access, but has added a controversial provision that these accepted manuscripts be licensed for free distribution. Some publishers have complained that this approach threatens their subscription revenues because it could widen free reading of these articles. Rooryck says Coalition S canvassed major publishers and found none was planning to routinely reject submitted manuscripts funded by Coalition S members because of the prospect that the authors would immediately post them when accepted. A spokesperson for publishing giant Elsevier told Science that all its journals will offer authors funded by Coalition S members the option to publish open access for a fee, allowing authors to comply with Plan S without violating embargoes. Where a researcher works strongly influences how much money is available for open-access fees. In Europe, institutions used dedicated internal funds to pay fees for 50% of articles their authors published in hybrid journals (those that publish both open-access and subscription content), but in the rest of the world, the figure was only 25%, according to a 2020 survey of authors by Springer Nature. Authors also tap funders and other sources, including their own personal funds. European scholars reported paying out of their own wallets for just 1% of the articles, compared with 16% in other countries. In Italy, the Nature group's new €9500 open-access fee has riled some researchers. That figure is “insane, there's no way on Earth to justify that,” says Manlio De Domenico, who leads a network science lab at the Bruno Kessler Foundation. The annual research budget for his 10-person lab recently included a total of €8000 for open-access fees for the entire year. “We can spend the money better another way,” he says—to pay Ph.D. students and, in normal times, fund travel to conferences and other labs. “To me, the trade-off is clear.” (The Nature group says the price reflects its costs to produce such highly selective journals; journals don't normally collect fees for papers they review but don't publish.) Nor do open-access publication fees hew closely to the laws of demand. One would expect fees to increase with the prestige of the journal, but a recent study by Schönfelder suggests that's not always true. She examined the relationship between fees paid by U.K. funders and the impact factor—a measure based on the average number of citations per article—of the journals where the papers appeared. She found a strong correlation in journals that published only open-access articles but a weaker one in hybrid journals. Hybrid journals tended to cost more than purely open-access journals, too. In a paper published last year, Schönfelder suggested her findings reflect the legacy of the subscription prices of large, traditional publishers such as Elsevier and Springer Nature, which publish many hybrid journals. These highly profitable companies with large shares of the publishing market have operated with limited competitive pressure. “If [their] pricing behavior wins through, the open-access transformation will come at a much higher cost than expected today,” Schönfelder wrote. A complete shift to open access could lead publishers to boost publishing fees even further, to try to make up for lost subscription revenues, says Claudio Aspesi, a publishing industry consultant based in Switzerland. Although just over 30% of all papers published in 2019 were paid open access, subscriptions still accounted for more than 90% of publishers' revenues that year, according to Delta Think, a consulting and marketing firm. Coalition S seeks to exert downward pressure on prices by increasing transparency. When a grantee's research is published, Plan S requires publishers to disclose to funders the basis for their prices, including the cost of services such as proofreading, copy editing, and organizing peer review. Rooryck says the coalition will share the information with authors and libraries, many of which help fund publishing fees. He expects the practice will increase price competition or provide “at a minimum, confidence that some of these prices are fair.” Despite wide acknowledgment by scientists, publishers, librarians, and policymakers of open access' potential benefits, many are reluctant to go all in. Even in Europe, where the movement for open access has been especially strong, Plan S is unusual. Of 60 funders surveyed in 2019, only 37 had an open-access policy, and only 23 monitored compliance, according to a report prepared for SPARC Europe, a nonprofit that advocates for open access. Some authors remain hesitant, too. In multiple surveys, authors have ranked open-access publishing below their need to publish in prestigious, high-impact journals to gain tenure and promotion. And they may be wary of a perception among some scientists that journals that carry only gold open-access articles lack rigor. (That view, researchers say, may reflect that such journals are relatively new, which lowers their impact factor.) A recent study also hints at inequities, finding that established, funded researchers at prestigious institutions are more likely to pay to publish their work open access. Anthony Olejniczak and Molly Wilson of the Academic Analytics Research Center, part of a data firm in Columbus, Ohio, examined the demographics and publishing patterns of more than 180,000 U.S. scholars. Overall, 84% of biological scientists and 66% in the physical and mathematical sciences had authored or co-authored at least one gold open-access paper between 2014 and 2018. Those authors were more likely to have advanced faculty rank and federal grants and to work at one of the 65 leading research universities that belong to the Association of American Universities, Olejniczak and Wilson report in an upcoming paper in Quantitative Science Studies . Olejniczak and Wilson hypothesize that scientists who choose to pay for open access not only need financial resources, but also the sense of job security that tenure confers. “This is a good news, bad news story,” Olejniczak says. “Open access is thriving, and it's growing.” But, he adds, publishers collecting the fees should consider ways to accommodate a wider diversity of authors. ![Figure][1] The many colors of open accessGRAPHIC: V. ALTOUNIAN/ SCIENCE One tenet of the open-access movement has been that publishing fees can be funded by redirecting money university libraries currently spend on journal subscriptions—but that assumption faces questions. Although the “transformative” agreements that cover both reading and publishing of articles have rapidly increased the percentage of articles published open access at some institutions, the details of these deals (like traditional, subscription-only ones) are often secret and have other features that make it difficult to compare bottom-line costs. Comparing costs across institutions is also challenging because these deals usually involve large packages of journals, with the exact lineup varying by institution. Still, it is clear that making most articles gold open access could wallop the library budgets of research-intensive universities whose scientists publish the most papers. Many institutions that publish little research would save money by dropping subscriptions and letting faculty members read articles for free, analysts say, and publishers would look to recoup the lost revenue through publishing fees. Pay It Forward , a report published by librarians at the University of California (UC) and colleagues in 2016, remains one of the most comprehensive analyses of the impact of these shifts on universities. They calculated what each of UC's 10 campuses and three comparison institutions would have paid to publish as gold open access all articles from between 2009 and 2013 that listed one of their faculty members as a corresponding author. A key finding: At most of the research-intensive institutions studied—such as the UC campuses in Los Angeles and San Francisco and Harvard University—simply redirecting funds from journal subscriptions wouldn't cover the open-access fees. Those institutions could charge the difference to federal grants, but they would still have to cover fees on papers from studies done without grant funding. Harvard, for example, might have to boost its total library spending by 71%, or nearly $6 million. Rich universities like Harvard could potentially tap their huge endowments and copious research funding to cover these costs, but other universities could struggle. U.S. university library budgets have lagged the rate of inflation in higher education for years and now face cuts because of the coronavirus pandemic. Some researchers interviewed for UC's study said they were reluctant to spend grant money on open-access publishing fees because they would eat into funds for research. “But in practice, we found [faculty members] are independently spending millions of dollars” from grants on fees, says MacKenzie Smith, university librarian at UC Davis and one of the study's co-authors. UC is conducting an experiment that limits the universities' contribution to per-article publication fees in order to encourage faculty members to consider other funding sources and journals with lower fees. “We want to get authors more engaged in the cost aspect of publishing, or at least mindful of it,” Smith says. If paying for open-access publication becomes the default route for scientists, and publishers hike prices as expected, many analysts worry publishing will become a luxury that only better funded researchers can afford. That could create a self-reinforcing cycle in which well-funded researchers publish more, potentially attracting more attention—and more funding. If that comes to pass, it could be especially hard on early-career researchers and authors in the developing world who lack their own grants, and on those in disciplines that traditionally receive less funding, such as math. Although publishers offer waivers for authors, many do not always cover the entire publishing fee or disclose what percentage of requests they grant. Small, nonprofit societies that currently depend on subscription fees from their journals could also lose out in an open-access world, because the dynamics of the pay-to-publish model tend to favor publishers and journals that produce a high volume of articles, which affords economies of scale. “I am worried that in the zeal to go that last mile” to make a larger portion of articles open access, “we could end up really hurting the scientific enterprise,” says Sudip Parikh, CEO of AAAS, which publishes the Science family of journals. One of them, Science Advances , charges an open-access fee of $4500, whereas the rest operate on the traditional subscription-only model. Parikh says AAAS is considering other options to make papers free to read, but he wasn't ready to discuss them when Science went to press. “I don't pretend to know the answer yet,” he says. “But it feels like there are other possibilities” besides publishing fees. One model for sustaining open access without relying on per-article publishing fees comes from Latin America. Brazil and other countries have funded the creation of free open-access journals and article repositories, and the region in 2019 had the world's highest percentage of scholarly articles available open access, 61%, according to the Curtin Open Knowledge Initiative. Debate continues about how to control publishing costs. Many advocates for open access say making it more affordable will require a vast shift in the culture of science. In particular, tenure and promotion committees will need to lower their expectations that authors publish in prestigious, costly journals. But some argue that even if funders and institutions must cough up more money to help authors publish open access, the potential to accelerate scientific discovery would justify the added cost. The journal publishing industry's annual revenues of about $10 billion represent less than 1% of total global spending on R&D—and, in this view, it's reasonable to divert more of the total to scholarly communications that are essential to making the entire enterprise run. It's unlikely, though, that all scientific articles will ever become open access, says Rick Anderson, university librarian at Brigham Young University, who has written extensively about business models for journal publishing. “It just seems to me like the barriers to universal open access are too great,” he says. “Every open-access model solves some problems and creates other problems.” “What I think is much more likely in the future, almost inevitable, is a fairly diverse landscape of open-access and subscription models,” Anderson adds. “I haven't yet seen anything that has convinced me that toll [subscription-based] access is going to go away entirely.” [1]: pending:yes