SCI COMMUN### Leadership President Joe Biden last week nominated former Senator Bill Nelson (D–FL) to lead NASA, a job that would require him to balance returning humans to the Moon with sending robots to observe Earth, Mars, and beyond. Nelson, who flew as a civilian on the space shuttle in 1986, was a lead Senate sponsor of NASA's new Moon rocket, the Space Launch System, which last week had the first successful test of its engines. Some advocates have called for replacing the delayed project, which has spent $20 billion so far, with privately operated rockets; but Nelson's nomination appears likely to cement the program. He has little track record of engagement with the agency's science programs, which have enjoyed rising budgets in recent years, especially for planetary science. But many members of Congress support an ambitious project to recover rock samples drilled by the Perseverance rover on Mars. Nelson is also seen as a backer of climate research, and earth scientists hope he will help reverse a long-term drop in NASA's spending on that discipline. 50% —The increase in atmospheric carbon dioxide since preindustrial times. The gas, which contributes to global warming, crossed that threshold for the first time when its concentration reached 417 parts per million this month (U.K. Met Office). ### Biodiversity Ecologists involved in mapping life on Earth have taken the next step: predicting where to find the 85% of the planet's species waiting to be discovered. Worried that conservation efforts usually ignore these hidden species, Mario Moura, an ecologist at the Federal University of Paraèba in Brazil, created an interactive map showing biodiversity hot spots with the richest potential for hosting unknown, land-dwelling species of mammals, birds, reptiles, and amphibians. He and ecologist Walter Jetz of Yale University built it using data about known vertebrates' size, habitat, and nine other attributes that influence the likelihood of discovery at a particular location. For example, large mammals living near people are much more likely to be found than tiny frogs in a remote jungle. The map predicts that Brazil has more mystery species, 10% of all, than any other country. Colombia, Indonesia, and Madagascar account for 5% each, Moura and Jetz reported this week in Nature Ecology & Evolution . ### COVID-19 The U.S. Centers for Disease Control and Prevention (CDC) last week named as a “variant of concern” a spreading strain of the pandemic coronavirus identified in California in January. The variant, which comes in two slightly different forms called B.1.427 and B.1.429, becomes the fourth given that designation by CDC. A variant of concern may evade immunity, drugs, and diagnostic tests; make people sicker; and be more transmissible than the original virus. A recent preprint study indicated that the new variant flagged by CDC resists antibodies generated by vaccines and natural infection. In the past 30 days, the variant has accounted for some 61% of the samples sequenced in California and 65% of sequences in Nevada. Also last week, the U.S. Food and Drug Administration updated its emergency use authorization for the monoclonal antibody bamlanivimab to indicate that when used alone, it is unlikely to be effective against the new variant. ### Infectious diseases In countries that bear a heavy burden of tuberculosis (TB), diagnosis and treatment for the disease declined markedly in 2020 because health care facilities closed or diverted resources to COVID-19, a nonprofit group said last week. The Stop TB Partnership collected data from nine countries, seven of them in Asia, that account for 60% of the global TB burden. It found that TB diagnosis and treatment fell on average by 23%; at least 1 million fewer new cases were treated, the lowest level since 2008. In 2019, TB was the world's leading infectious killer, with 1.4 million deaths, more than 95% of them in developing countries. The Stop TB Partnership predicts that in 2020, TB remained the leading infectious killer in these countries even though COVID-19 surpassed it globally. The group urges health systems to simultaneously test for and tackle COVID-19 and TB, both infectious respiratory diseases. ### Climate change It's back! The U.S. Environmental Protection Agency (EPA) last week relaunched a web page focused on climate change that former President Donald Trump's administration dismantled 4 years ago. The agency says it plans to add new, updated information in coming months about climate science and its actions to curb global warming. In 2017, then-EPA Administrator Scott Pruitt erased the page, which had existed for some 20 years, along with many other mentions of climate science on the agency's website. ### Agriculture The golden apple snail ( Pomacea canaliculata ), an invasive pest that costs rice farmers in Asia more than $1 billion a year, has been documented in continental Africa for the first time. Farmers in Kenya's Mwea irrigation area, a 10,000-hectare region that produces more than 70% of the country's rice, noticed the 2-centimeter snails, which were later confirmed by DNA tests, researchers report this week in CABI Agriculture and Bioscience . The snails are eating 90% of seedlings, the farmers say. The snail species is native to the Americas. To help prevent its spread in Africa, researchers say Kenya should consider restricting movement of farm machinery and host plants from the infected area. ### Biomedicine The founding director of the U.S. National Institutes of Health's (NIH's) center focusing on the translation of basic science into treatments is stepping down. Neurologist and geneticist Christopher Austin, who led the National Center for Advancing Translational Sciences (NCATS), will become a partner at the venture capital firm Flagship Pioneering. NIH Director Francis Collins persuaded Congress to create NCATS in 2011, despite skepticism from industry. Under Austin, the center's efforts included screening small molecules for drug candidates, developing treatments for rare diseases, and streamlining clinical trials at academic health centers. Some observers say NCATS never had the budget to realize its ambitions ( Science , 27 September 2019, p. [1363][1]). But the government's role in drug development may soon grow, with President Joe Biden expected to reveal details for a health agency modeled after the applications-oriented Defense Advanced Research Projects Agency. ### Drug development A late-stage drug trial that had buoyed the hopes of families afflicted by Huntington disease, a devastating, incurable neurological illness, was stopped on 22 March by the manufacturer, Roche. A data monitoring committee had recommended the move after weighing benefits and risks of the drug, tominersen (formerly called IONIS-HTTRx and RG6042). The company said no new safety issues had emerged during the trial, but provided no additional details. Tominersen is an antisense oligonucleotide, a snippet of genetic code designed to interfere with production of a protein—in this case, the mutant protein called huntingtin that causes the inherited disease. The drug was developed by Ionis Pharmaceuticals, which licensed it to Roche in 2017, shortly after an early human trial showed it caused dose-dependent reductions in huntingtin ( Science , 24 August 2018, p. [742][2]). The sudden cessation stunned trial leaders, who had finished enrolling nearly 800 people in April 2020 and expected results in 2022. ![Figure][3] In Focus The stems, air bladders, and blades of giant kelp ( Macrocystis pyrifera ), the world's largest seaweed, form a golden still life. With fronds reaching more than 30 meters, the species thrives in cold, clear water, forming dense forests that host diverse ecosystems. This prize-winning image is part of an outdoor exhibit about algae at the University of Nottingham that can be seen at . PHOTO: ERASMO EMACAYA ### Animal behavior Adult mountain gorillas, including male silverbacks, often adopt motherless or orphaned young members of their social groups—a behavior previously thought to be rare in mammals other than humans—a study reports. Chimpanzees and other social mammals that lose their mothers face a greater risk of dying, and gorillas often kill orphans from different social groups. But thanks to the in-group communal care, gorillas who were motherless or orphaned survived as well as youngsters whose mothers were still around. This week in eLife , scientists describe the analysis, which used a remarkable 53 years of data on mountain gorillas collected at the Dian Fossey Gorilla Fund's Karisoke Research Center in Rwanda, in one of the longest field studies. Scientists found that 59 motherless or orphaned gorillas also suffered no long-term decline in social rank or ability to reproduce. Older sisters and peers of orphans joined males, including dominant silverbacks, in frequent child care. Some motherless males who spent lots of time with the silverback in their social group did well: They grew up to become its dominant ape. ### Climate science The Biden administration last week appointed Jane Lubchenco, who headed the National Oceanic and Atmospheric Administration under former President Barack Obama, to serve as its top climate scientist in the White House Office of Science and Technology Policy. A marine ecologist and advocate for the ocean, Lubchenco will work as deputy director for climate and the environment, coordinating climate science across federal agencies and the U.S. Global Change Research Program. Lubchenco's focus will be to spur research and action on climate mitigation that can also benefit the U.S. economy. ### Science policy In Sweden, a national code uses 44,000 words to define research misconduct and discuss scientific values. Next door, Norway's equivalent is a brisk 900 words. An analysis of scientific integrity policies in 32 European nations has found wide differences not only in length, but in standards and definitions of research misconduct. The nonbinding European Code of Conduct for Research Integrity, issued by the European Federation of Academies of Sciences and Humanities in 2017, was intended to promote consistency; it also allows countries to add language to fit their circumstances. But only two of 32 countries examined—Bulgaria and Luxembourg—have adopted the code wholesale, according to a study last month in Bioethics . And there's just one principle that all countries have agreed on: that fabrication, falsification, and plagiarism of data and findings constitute research misconduct. Research ethicists say the differences threaten to create confusion and disputes for multinational scientific collaborations as they decide which country's rules apply. ### Microbiology Soil bacteria called Streptomyces are guardian angels of the microbial world: They produce antibiotics that humans depend on and protect plants from harmful microbes. But because they lack self-propulsion, researchers have long puzzled over how they find the plants they defend. Now, scientists have discovered that the microbe's dormant spores typically hitch rides on the whiplike flagella of other, mobile soil microbes heading for plant roots, researchers reported last week in The ISME Journal: Multidisciplinary Journal of Microbial Ecology . A few other species of bacteria and a fungus are known to catch rides on other microbes. But the observed piggybacking on microbes by spores—previously known to latch on to insects and other small animals for travel—is a first, says co-author Alise Muok of Leiden University. With this mystery solved, researchers may one day modify the hitchhiking to improve plant protection, she adds. [1]: http://www.sciencemag.org/content/365/6460/1363 [2]: http://www.sciencemag.org/content/361/6404/742 [3]: pending:yes

We consider the following problem in which a given number of items has to be chosen from a predefined set. Each item is described by a vector of attributes and for each attribute there is a desired distribution that the selected set should have. We look for a set that fits as much as possible the desired distributions on all attributes. Examples of applications include choosing members of a representative committee, where candidates are described by attributes such as sex, age and profession, and where we look for a committee that for each attribute offers a certain representation, i.e., a single committee that contains a certain number of young and old people, certain number of men and women, certain number of people with different professions, etc. With a single attribute the problem collapses to the apportionment problem for party-list proportional representation systems (in such case the value of the single attribute would be a political affiliation of a candidate). We study the properties of the associated subset selection rules, as well as their computation complexity.

In this work, we propose a novel deep learning framework that can generate a vivid dance from a whole piece of music. In contrast to previous works that define the problem as generation of frames of motion state parameters, we formulate the task as a prediction of motion curves between key poses, which is inspired by the animation industry practice. The proposed framework, named DanceNet3D, first generates key poses on beats of the given music and then predicts the in-between motion curves. DanceNet3D adopts the encoder-decoder architecture and the adversarial schemes for training. The decoders in DanceNet3D are constructed on MoTrans, a transformer tailored for motion generation. In MoTrans we introduce the kinematic correlation by the Kinematic Chain Networks, and we also propose the Learned Local Attention module to take the temporal local correlation of human motion into consideration. Furthermore, we propose PhantomDance, the first large-scale dance dataset produced by professional animatiors, with accurate synchronization with music. Extensive experiments demonstrate that the proposed approach can generate fluent, elegant, performative and beat-synchronized 3D dances, which significantly surpasses previous works quantitatively and qualitatively.

Robotic manipulation of unknown objects is an important field of research. Practical applications occur in many real-world settings where robots need to interact with an unknown environment. We tackle the problem of reactive grasping by proposing a method for unknown object tracking, grasp point sampling and dynamic trajectory planning. Our object tracking method combines Siamese Networks with an Iterative Closest Point approach for pointcloud registration into a method for 6-DoF unknown object tracking. The method does not require further training and is robust to noise and occlusion. We propose a robotic manipulation system, which is able to grasp a wide variety of formerly unseen objects and is robust against object perturbations and inferior grasping points.

The unification of statistical (data-driven) and symbolic (knowledge-driven) methods is widely recognised as one of the key challenges of modern AI. Recent years have seen large number of publications on such hybrid neuro-symbolic AI systems. That rapidly growing literature is highly diverse and mostly empirical, and is lacking a unifying view of the large variety of these hybrid systems. In this paper we analyse a large body of recent literature and we propose a set of modular design patterns for such hybrid, neuro-symbolic systems. We are able to describe the architecture of a very large number of hybrid systems by composing only a small set of elementary patterns as building blocks. The main contributions of this paper are: 1) a taxonomically organised vocabulary to describe both processes and data structures used in hybrid systems; 2) a set of 15+ design patterns for hybrid AI systems, organised in a set of elementary patterns and a set of compositional patterns; 3) an application of these design patterns in two realistic use-cases for hybrid AI systems. Our patterns reveal similarities between systems that were not recognised until now. Finally, our design patterns extend and refine Kautz' earlier attempt at categorising neuro-symbolic architectures.

The world of AI is rapidly changing, and we'd like to help you keep up with it. We've curated a list of the Top 20 AI thought leaders that you should follow in 2021. This list comprises global business leaders, authors, AI advocates, data scientists, and professors. Follow them, reach out to them, and see how they can guide you on your AI journey. Each of these AI leaders has deep insights that will help you optimize your business with artificial intelligence.

In our JournalismAI report, journalists around the world told researchers they are eager to collaborate and explore the benefits of AI, especially as it applies to newsgathering, production and distribution. To facilitate their collaboration, the Google News Initiative and Polis – the journalism think tank at the London School of Economics and Political Science – are launching the JournalismAI Collab Challenges, an opportunity for three groups of five newsrooms from the Americas, Europe, the Middle East and Africa, and Asia Pacific to experiment together. Each cohort – selected by Polis – will have six months to either cover global news stories using AI-powered storytelling techniques or to develop prototypes of new AI-based products and processes. Participants will receive support from the JournalismAI team and partner institutions in each region: in the Americas, the challenge will be co-hosted with the Knight Lab at Northwestern University; in Europe, the Middle East and Africa, the challenge will be co-hosted with BBC News Labs and Clwstwr. JournalismAI's partner in Asia Pacific will be announced later this year.

Farmers have long struggled with operational optimization and labor concerns. Finding enough labor to get the job done, as well as keeping workers safe is a constant struggle. "There is an immediate need to improve efficiency and reduce costs, especially now that the pandemic has exposed just how fragile the supply chain is," said Suma Reddy, CEO of Future Acres, an agricultural robotics and artificial intelligence company. "We saw shortages in both production and more workers being put at risk when picking specialty crops on a daily basis that have really caused the industry to take a step back and re-examine how we can create greater resiliency in the food chain." One idea is to equip farms with a combination of AI and robotics that can "think through" as well as do some of the physical work of farming.

Text-based games simulate worlds and interact with players using natural language. Recent work has used them as a testbed for autonomous language-understanding agents, with the motivation being that understanding the meanings of words or semantics is a key component of how humans understand, reason, and act in these worlds. However, it remains unclear to what extent artificial agents utilize semantic understanding of the text. To this end, we perform experiments to systematically reduce the amount of semantic information available to a learning agent. Surprisingly, we find that an agent is capable of achieving high scores even in the complete absence of language semantics, indicating that the currently popular experimental setup and models may be poorly designed to understand and leverage game texts. To remedy this deficiency, we propose an inverse dynamics decoder to regularize the representation space and encourage exploration, which shows improved performance on several games including Zork I. We discuss the implications of our findings for designing future agents with stronger semantic understanding.

Learning continuously during all model lifetime is fundamental to deploy machine learning solutions robust to drifts in the data distribution. Advances in Continual Learning (CL) with recurrent neural networks could pave the way to a large number of applications where incoming data is non stationary, like natural language processing and robotics. However, the existing body of work on the topic is still fragmented, with approaches which are application-specific and whose assessment is based on heterogeneous learning protocols and datasets. In this paper, we organize the literature on CL for sequential data processing by providing a categorization of the contributions and a review of the benchmarks. We propose two new benchmarks for CL with sequential data based on existing datasets, whose characteristics resemble real-world applications. We also provide a broad empirical evaluation of CL and Recurrent Neural Networks in class-incremental scenario, by testing their ability to mitigate forgetting with a number of different strategies which are not specific to sequential data processing. Our results highlight the key role played by the sequence length and the importance of a clear specification of the CL scenario.