Government
Yes We Care! -- Certification for Machine Learning Methods through the Care Label Framework
Morik, Katharina, Kotthaus, Helena, Heppe, Lukas, Heinrich, Danny, Fischer, Raphael, Mücke, Sascha, Pauly, Andreas, Jakobs, Matthias, Piatkowski, Nico
Machine learning applications have become ubiquitous. Their applications from machine embedded control in production over process optimization in diverse areas (e.g., traffic, finance, sciences) to direct user interactions like advertising and recommendations. This has led to an increased effort of making machine learning trustworthy. Explainable and fair AI have already matured. They address knowledgeable users and application engineers. However, there are users that want to deploy a learned model in a similar way as their washing machine. These stakeholders do not want to spend time understanding the model. Instead, they want to rely on guaranteed properties. What are the relevant properties? How can they be expressed to stakeholders without presupposing machine learning knowledge? How can they be guaranteed for a certain implementation of a model? These questions move far beyond the current state-of-the-art and we want to address them here. We propose a unified framework that certifies learning methods via care labels. They are easy to understand and draw inspiration from well-known certificates like textile labels or property cards of electronic devices. Our framework considers both, the machine learning theory and a given implementation. We test the implementation's compliance with theoretical properties and bounds. In this paper, we illustrate care labels by a prototype implementation of a certification suite for a selection of probabilistic graphical models.
From parcel to continental scale -- A first European crop type map based on Sentinel-1 and LUCAS Copernicus in-situ observations
d'Andrimont, Raphaël, Verhegghen, Astrid, Lemoine, Guido, Kempeneers, Pieter, Meroni, Michele, van der Velde, Marijn
Detailed parcel-level crop type mapping for the whole European Union (EU) is necessary for the evaluation of agricultural policies. The Copernicus program, and Sentinel-1 (S1) in particular, offers the opportunity to monitor agricultural land at a continental scale and in a timely manner. However, so far the potential of S1 has not been explored at such a scale. Capitalizing on the unique LUCAS 2018 Copernicus in-situ survey, we present the first continental crop type map at 10-m spatial resolution for the EU based on S1A and S1B Synthetic Aperture Radar observations for the year 2018. Random forest classification algorithms are tuned to detect 19 different crop types. We assess the accuracy of this EU crop map with three approaches. First, the accuracy is assessed with independent LUCAS core in-situ observations over the continent. Second, an accuracy assessment is done specifically for main crop types from farmers declarations from 6 EU member countries or regions totaling >3M parcels and 8.21 Mha. Finally, the crop areas derived by classification are compared to the subnational (NUTS 2) area statistics reported by Eurostat. The overall accuracy for the map is reported as 80.3% when grouping main crop classes and 76% when considering all 19 crop type classes separately. Highest accuracies are obtained for rape and turnip rape with user and produced accuracies higher than 96%. The correlation between the remotely sensed estimated and Eurostat reported crop area ranges from 0.93 (potatoes) to 0.99 (rape and turnip rape). Finally, we discuss how the framework presented here can underpin the operational delivery of in-season high-resolution based crop mapping.
Automate feature engineering pipelines with Amazon SageMaker
The process of extracting, cleaning, manipulating, and encoding data from raw sources and preparing it to be consumed by machine learning (ML) algorithms is an important, expensive, and time-consuming part of data science. Managing these data pipelines for either training or inference is a challenge for data science teams, however, and can take valuable time away that could be better used towards experimenting with new features or optimizing model performance with different algorithms or hyperparameter tuning. Many ML use cases such as churn prediction, fraud detection, or predictive maintenance rely on models trained from historical datasets that build up over time. The set of feature engineering steps a data scientist defined and performed on historical data for one time period needs to be applied towards any new data after that period, as models trained from historic features need to make predictions on features derived from the new data. Instead of manually performing these feature transformations on new data as it arrives, data scientists can create a data preprocessing pipeline to perform the desired set of feature engineering steps that runs automatically whenever new raw data is available.
Teaching artificial intelligence to control navy submarines
In the future Cdr Ramsey believes submarines will be equipped with the ability to launch their own autonomous vessels. The next generation of surface warships entering service are already planned to feature mission-bays for launching uncrewed vessels. And autonomous submarines can be deployed by transport aircraft, giving the batteries a rest and the XLUUV a global reach.
A Sense Of Touch Boosts Speed, Accuracy Of Mind-Controlled Robotic Arm
President Barack Obama bumped fists with Nathan Copeland during a tour of innovation projects at the White House Frontiers Conference at the University of Pittsburgh in 2016. President Barack Obama bumped fists with Nathan Copeland during a tour of innovation projects at the White House Frontiers Conference at the University of Pittsburgh in 2016. A robotic arm with a sense of touch has allowed a man who is paralyzed to quickly perform tasks like pouring water from one cup into another. The robotic arm provides tactile feedback directly to the man's brain as he uses his thoughts to control the device, a team reports Thursday in the journal Science. Previous versions of the arm required the participant, Nathan Copeland, to guide the arm using vision alone.
Drones Flying Near Airports, Infrastructure Prompt U.S. Action
Federal agencies are scrambling to address a surge in the use of consumer drones as the unmanned aircraft crowd the airspace above critical sites, posing a threat to public safety and national security. The Federal Aviation Administration and the National Aeronautics and Space Administration are developing a joint national air-traffic-control system for low-flying drones. The Department of Homeland Security is testing technologies to detect small drones favored by consumers, and the Pentagon is researching methods to knock them out of the sky. Reports of drone sightings around airports are pouring into the FAA at a rate of more than 100 a month. Commercial pilots flying into and out of Los Angeles International Airport have reported increased sightings of drones near their flight path, with 23 sightings reported to the control tower so far this year, according to an airport official.
NASA's Ingenuity helicopter to attempt its SIXTH flight on Mars next week
NASA's Ingenuity helicopter is going to attempt a sixth flight on Mars next week, reaching speeds of up to nine miles per hour and flying for over two minutes. This will be the first flight as part of the'operations demonstration phase' that includes helping the Perseverance rover in its mission to find ancient signs of life. Ingenuity will scout multiple surface features from the air during its short flight, before landing in a different airfield to the one it takes off from for the first time. It will head up 33 feet into the Martian sky and then fly 492ft southwest at about nine miles per hour, taking photographs of rocks and dunes on the ground. In this new phase, data and images from the flight will be returned to Earth in the days following the flight, rather than in the following hours as with earlier trips. NASA's Ingenuity helicopter is going to attempt a sixth flight on Mars next week, reaching speeds of up to nine miles per hour and flying for over two minutes This will be the first flight as part of the'operations demonstration phase' that includes helping the Perseverance rover in its mission to find ancient signs of life Ingenuity was designed as a technology demonstrator rather than carrying any of its own science experiments or equipment.
A psychedelic win raises questions for PTSD therapy
The news last week that the compound 3,4-methylenedioxymethamphetamine (MDMA), popularly called ecstasy, alleviated post-traumatic stress disorder (PTSD) in a phase 3 trial was a milestone in efforts to turn psychedelic drugs into mainstream treatments. It also highlighted a therapeutic marriage that is getting increasing attention: providing a mind-altering drug while a patient receives care from a trained therapist. “This is really kind of a new zeitgeist in psychiatry,” says Barbara Rothbaum, a clinical psychologist at Emory University. Researchers funded by the nonprofit Multidisciplinary Association for Psychedelic Studies (MAPS) reported in Nature Medicine that after “MDMA-assisted therapy,” two-thirds of treated participants no longer met the diagnostic criteria for PTSD. MAPS hopes to confirm the results in an ongoing second study and seek approval for the therapy from the U.S. Food and Drug Administration (FDA) as early as 2023. In 2017, the agency granted MDMA a “breakthrough” designation, which comes with extra guidance during the trial process and an expedited review. The idea that psychedelic drugs and talk therapy work in synergy raises complex questions about how to optimize—and regulate—the drug experience. “It's not understood what the role of MDMA or [another] psychedelic is in facilitating the psychotherapy, and what's happening neurobiologically,” says Atheir Abbas, a neuroscientist and psychiatrist at Oregon Health & Science University. Because taking psychedelics without guidance can lead to negative experiences, “a guided, more psychotherapy-oriented approach is probably warranted,” he says. “But it's not clear what aspects of that guidance are critical.” As MDMA and other tightly controlled psychedelic compounds inch closer to regulatory approval, careful supervision from therapists may help overcome their reputation as illicit substances and fears of indiscriminate use. “MAPS has done very well to really emphasize that they're not just trying to promote a compound,” says Rachel Yehuda, who directs the Center for Psychedelic Psychotherapy and Trauma Research at the Icahn School of Medicine at Mount Sinai. MAPS researchers have worked for more than 3 decades to turn MDMA into a prescription medicine. The substance doesn't produce the vivid hallucinations associated with either LSD or psilocybin, found in magic mushrooms. But it increases the brain's levels of certain neurotransmitters, including serotonin and dopamine, to create a sense of well-being and heightened empathy. That might allow trauma survivors who face intrusive flashbacks to reflect on disturbing memories with less fear and judgement. “It gives you this fascinating ability toward self-compassion,” says Jennifer Mitchell, a neuroscientist at the University of California, San Francisco, and an investigator in the MAPS trial. The psychotherapy the trial used looks different from the most thoroughly studied PTSD psychotherapies. Many of these direct a patient to confront painful memories. In one such approach, called prolonged exposure therapy, “we have them go back in their mind's eye to the time of the traumatic event and recount it out loud, in the present tense, over and over,” Rothbaum says. “It's a good therapy, but it's also a hard therapy.” In the MAPS approach, therapists are advised not to steer the conversation toward trauma, but to create a safe space to “support the participant's own unfolding experience,” according to MAPS's manual. The manual also describes carefully curated surroundings for patients, advising “fresh flowers and artwork.” Rothbaum, who studies and practices prolonged exposure therapy, describes MAPS sessions as “the sweetest therapy in the world for PTSD.” In the trial, 79 participants underwent three 90-minute preparatory therapy sessions; three 8-hour “experimental sessions” with either MDMA or a placebo, spaced about 1 month apart; and nine 90-minute “integration sessions” to process their experiences. Of 42 people who got MDMA, 67% no longer met the diagnostic criteria for PTSD 2 months after their last experimental session. For the placebo group, that rate was 32%. Improvements in the placebo recipients were comparable to those observed in studies of existing approaches such as exposure therapy, says MAPS's chief scientific officer, Berra Yazar-Klosinski. Yehuda, who has long worried that prolonged exposure therapy can retraumatize some patients and discourage them from continuing treatment, says the research offers a new direction. But MDMA-assisted therapy needs to go head to head in clinical trials with established psychotherapies and the antidepressant drugs already approved for PTSD, says Arash Javanbakht, who directs the Stress, Trauma, and Anxiety Research Clinic at Wayne State University. Others want to test MDMA alongside more mainstream talk therapies. Rothbaum's team hopes to launch a pilot study this year that combines the drug with prolonged exposure therapy. Psychologist Anne Wagner of the mental health clinic Remedy is preparing to launch a pilot study that brings an approach called cognitive processing therapy into the preand post-MDMA sessions. Adding this more structured psychotherapy might improve results for some patients, she says. Should MDMA win FDA approval, the MAPS psychotherapy approach is likely to dominate at least at first. FDA doesn't regulate psychotherapy, but Yazar-Klosinski says FDA has indicated any approval would stipulate that the drug be used alongside therapy from trained providers. (FDA declined to comment on how it might address supportive therapy in an approval.) A subsidiary of MAPS would manufacture the drug and hopes to receive a 5-year exclusivity period on marketing the drug for PTSD. It would sell MDMA only to providers who undergo MAPS's 100-hour training program. Uncertainty about the right talk therapy pairings extends to research on other psychedelic drugs. Albert Garcia-Romeu, a psychologist at Johns Hopkins University School of Medicine, is testing psilocybin-assisted addiction treatment. For a recent study of cigarette smokers, the team made “kind of our best guess” to pair the drug with cognitive behavioral therapy, a common addiction talk therapy, he says. Potential FDA decisions on psilocybin and MDMA are still years off. But in November 2020, Oregon became the first state to legalize psilocybin for therapeutic use. It has appointed an advisory board to help make a plan for regulating the compound by December 2022. Oregon's measure describes a system of state-licensed providers and “a multisession, facilitator-guided, psychotherapy-type experience,” says Abbas, who sits on the advisory board but was speaking only for himself. What type of psychotherapy experience should that be? And what qualifies a facilitator? “That's something the board is likely to have to tackle.”
The Space Force Says We Have to Try Human Augmentation
The chief scientist for the U.S. Space Force says human technological augmentation won't be optional in the future, and the U.S. will need to invest in it to keep pace with countries like Russia and China. Rather than radical redesigns of the human body, however, Joel Mozer proposes augmenting humans with artificial intelligence, including AI so advanced, it can advise commanders on tactics and keep track of complex problems. Mozer, a member of the Air Force Research Lab (AFRL), made the comments during an AFRL live event streamed on Facebook. "Today, we are on the brink of a new age, the age of human augmentation," Mozer said. "It's imperative that we embrace this new age, lest we fall behind our strategic competitors."
New law of physics could improve grip in robotics
Researchers at North Carolina State University have discovered a new law of physics that could enable improved friction and grip in robotics. It can be difficult for engineers to account for the friction that occurs when robots grip objects, particularly in wet environments. This is due to elastohydrodynamic lubrication (EHL), the friction that occurs when two solid surfaces come into contact with a thin layer of fluid between them. In humans, this friction occurs when fingertips are rubbed together, the fluid being the thin layer of naturally occurring oil on the skin. This could also apply to a robotic claw lifting an object that has been coated with oil, or to a surgical device being used inside the human body.