Tyler O'Neal, Staff Editor CLOUD

Duke's machine learning shapes microwaves for a computer's eyes

Hardware and software tweak microwave patterns to discover the most efficient way to identify objects

Engineers from Duke University and the Institut de Physique de Nice in France have developed a new method to identify objects using microwaves that improves accuracy while reducing the associated computing time and power requirements.

The system could provide a boost to object identification and speed in fields where both are critical, such as autonomous vehicles, security screening, and motion sensing.

The new machine-learning approach cuts out the middleman, skipping the step of creating an image for analysis by a human and instead analyzes the pure data directly. It also jointly determines optimal hardware settings that reveal the most important data while simultaneously discovering what the most important data is. In a proof-of-principle study, the setup correctly identified a set of 3D numbers using tens of measurements instead of the hundreds or thousands typically required.  CAPTION An example of a wave pattern (right) and its intensity levels (left) developed by the machine learning algorithm to best illuminate the most important features of an object being identified. CREDIT Mohammadreza Imani, Duke University{module INSIDE STORY}

The results appear online on December 6 in the journal Advanced Science and are a collaboration between David R. Smith, the James B. Duke Distinguished Professor of Electrical and Computer Engineering at Duke, and Roarke Horstmeyer, assistant professor of biomedical engineering at Duke.

"Object identification schemes typically take measurements and go to all this trouble to make an image for people to look at and appreciate," said Horstmeyer. "But that's inefficient because the computer doesn't need to 'look' at an image at all."

"This approach circumvents that step and allows the program to capture details that an image-forming process might miss while ignoring other details of the scene that it doesn't need," added Aaron Diebold, a research assistant in Smith's lab. "We're basically trying to see the object directly from the eyes of the machine."

In the study, the researchers use a metamaterial antenna that can sculpt a microwave wavefront into many different shapes. In this case, the metamaterial is an 8x8 grid of squares, each of which contains electronic structures that allow it to be dynamically tuned to either block or transmit microwaves.

For each measurement, the intelligent sensor selects a handful of squares to let microwaves pass through. This creates a unique microwave pattern, which bounces off the object to be recognized and returns to another similar metamaterial antenna. The sensing antenna also uses a pattern of active squares to add further options to shape the reflected waves. The computer then analyzes the incoming signal and attempts to identify the object.

By repeating this process thousands of times for different variations, the machine learning algorithm eventually discovers which pieces of information are the most important as well as which settings on both the sending and receiving antennas are the best at gathering them.

"The transmitter and receiver act together and are designed together by the machine learning algorithm," said Mohammadreza Imani, a research assistant in Smith's lab. "They are jointly designed and optimized to capture the features relevant to the task at hand."

"If you know your task, and you know what sort of scene to expect, you may not need to capture all the information possible," said Philipp del Hougne, a postdoctoral fellow at the Institut de Physique de Nice. "This co-design of measurement and processing allows us to make use of all the a priori knowledge that we have about the task, scene and measurement constraints to optimize the entire sensing process."

After training, the machine learning algorithm landed on a small group of settings that could help it separate the data's wheat from the chaff, cutting down on the number of measurements, time and computational power it needs. Instead of the hundreds or even thousands of measurements typically required by traditional microwave imaging systems, it could see the object in less than 10 measurements.  CAPTION In a new type of object identification, a radio wave source (back panel) creates a wave front (middle panel) that is shaped by a metamaterial screen which allows waves to pass through in some places but not others (front panel). Machine learning then finds the wave shapes that illuminate the most useful features of an object. The method improves accuracy while reducing computing time and power requirements. CREDIT Mohammadreza Imani, Duke University{module INSIDE STORY}

Whether or not this level of improvement would scale up to more complicated sensing applications is an open question. But the researchers are already trying to use their new concept to optimize hand-motion and gesture recognition for next-generation computer interfaces. There are plenty of other domains where improvements in microwave sensing are needed, and the small size, low cost and easy manufacturability of these types of metamaterials make them promising candidates for future devices.

"Microwaves are ideal for applications like concealed threat detection, identifying objects on the road for driverless cars or monitoring for emergencies in assisted-living facilities," said del Hougne. "When you think about all of these applications, you need the sensing to be as quick as possible, so we hope our approach will prove useful in making these ideas reliable realities."

York's mathematicians put famous Battle of Britain 'what if' scenarios to the test

Tyler O'Neal, Staff Editor CLOUD

Mathematicians have used a statistical technique to interrogate some of the big "what if" questions in the Second World War battle for Britain's skies.

What if the switch to bombing London had not occurred? What if a more eager Hitler had pushed for an earlier beginning to the campaign? What if Goring had focused on targeting British airfields throughout the entire period of the Battle?

These are just some of the alternative scenarios that have formed a long-running debate among Second World War historians and enthusiasts over what might have affected the outcome of the battle, which took place between May and October 1940.

Mathematicians from the University of York have developed a new model to explore what the impact of changes to Luftwaffe tactics would have been. Their approach uses statistical modeling to calculate how the Battle might have played out if history had followed one of several alternative courses.

The researchers say that the method could now be used as a tool to investigate other historical controversies and unrealized possibilities, giving us a deeper understanding of events such as the Battle of the Atlantic (the longest continuous military campaign of the Second World War).

The statistical technique is called "weighted bootstrapping" and the supercomputer simulation is a bit like taking a ball for the events of each day of the Battle of Britain and placing them in a lotto machine. Balls are then drawn, read and replaced to create thousands of alternative sets of days' fighting, but in a different order, and perhaps with some days appearing more than once or not at all.

The researchers then repeated the process to test out the Battle "what ifs", making some days more or less likely to be chosen, depending on how a protagonist (such as Hitler) would have changed their decisions had they been using different tactics.

Co-author of the paper, Dr. Jamie Wood from the Department of Mathematics at the University of York, said: "The weighted bootstrap technique allowed us to model alternative campaigns in which the Luftwaffe prolongs or contracts the different phases of the battle and varies its targets.  {module INSIDE STORY}

"The Luftwaffe would only have been able to make the necessary bases in France available to launch an air attack on Britain in June at the earliest, so our alternative campaign brings forward the air campaign by three weeks. We tested the impact of this and the other counterfactuals by varying the probabilities with which we choose individual days."

The results provide statistical backing to a change in tactics that several historians have argued could have brought the Luftwaffe victory in the summer of 1940: The simulations suggested that if they had started the campaign earlier and focused on bombing airfields, the RAF might have been defeated, paving the way for a German land invasion.

According to the mathematical model, the impact of these two changes would have been dramatic. Although it is impossible to estimate what the real statistical chances of an RAF victory were in July 1940, the study suggests that whatever Britain's prospects, an earlier start and focused targeting of airfields would have shifted the battle significantly in the Germans' favor.

For example, had the likelihood of a British victory in the actual battle been 50%, these two tactical changes would have reduced it to less than 10%. If the real probability of British victory was 98%, the same changes would have reduced this to just 34%.

Co-author of the paper, Professor Niall Mackay from the Department of Mathematics at the University of York, said: "Weighted bootstrapping can provide a natural and intuitive tool for historians to investigate unrealized possibilities, informing historical controversies and debates.

"It demonstrates just how finely-balanced the outcomes of some of the biggest moments of history were. Even when we use the actual days' events of the battle, make a small change of timing or emphasis to the arrangement of those days and things might have turned out very differently.

"This technique can be used to give us a more complete understanding of just how different events might have played out."

Million Veteran Program study sheds light on genetic basis of anxiety

Tyler O'Neal, Staff Editor CLOUD

Largest genetic study of anxiety to date reports findings

A massive genome-wide analysis of approximately 200,000 military veterans has identified six genetic variants linked to anxiety, researchers from Yale and colleagues at other institutions report Jan. 7 in the American Journal of Psychiatry.

The study used VA Million Veteran Program (MVP) data to identify regions on the human genome related to anxiety risk. The findings could lead to new understanding and treatment of the condition, which affects 1 in 10 Americans.

Some of the variants associated with anxiety had previously been implicated as risk factors for bipolar disorder, posttraumatic stress disorder, and schizophrenia.

According to Dr. Dan Levey of the VA Connecticut Healthcare Center and Yale University, one of the lead authors on the study, the findings are "an important step forward" in the understanding of anxiety disorders and how genes contribute to mental conditions.

Anxiety refers to the anticipation of perceived future threats. In anxiety disorders, these concerns are out of proportion to the actual anticipated event, leading to distress and disability. Anxiety disorders often occur alongside other mental health disorders like depression. {module INSIDE STORY}

Only a third of those with anxiety disorders receive treatment. Some forms of psychotherapy, such as cognitive-behavioral therapy, have proved effective, as have medications such as selective serotonin reuptake inhibitors. In other fields of medicine, genetic studies have led to precision medicine approaches--tailoring drug treatment to patients' individual genetic and biochemical profiles--for a number of diseases. The researchers hope more genetic insight will lead to similar approaches for anxiety.

The findings for the African American participants are especially important, says Levey. "Minorities are underrepresented in genetic studies, and the diversity of the Million Veteran Program was essential for this part of the project. The genetic variant we identified occurs only in individuals of African ancestry, and would have been completely missed in less diverse cohorts."

The study produced the first genome-wide significant findings on anxiety in African ancestry, notes Levey. About 18% of MVP participants are African American.

"This is the richest set of results for the genetic basis of anxiety to date," said co-lead author Joel Gelernter, the Foundations Fund Professor of Psychiatry, professor of genetics and of neuroscience at Yale. "There has been no explanation for the comorbidity of anxiety and depression and other mental health disorders, but here we have found specific, shared genetic risks."

Finding the genetic underpinnings of mental health disorders is the primary goal of the Million Veteran Program, a compilation of health and genetic data on U.S. military veterans run by the U.S. Veterans Administration. The research team analyzed the program's data and zeroed in on six variants linked to anxiety. Five were found in European Americans and one found only in African Americans. Gene variants at these genome locations could increase anxiety risk, say the scientists.

The anxiety-related genome locations also show overlap with other psychiatric conditions. One of the identified locations has previously been linked with risk for bipolar disorder and schizophrenia. The study also shows a genetic overlap between anxiety symptoms and depression, PTSD (which is related to anxiety), and neuroticism--a personality trait that has been shown to increase the risk for anxiety and related disorders. The results support the idea that overlap with these other traits is at least partially due to a significant genetic commonality, according to the researchers.

MVP is a national, voluntary research program funded by VA's Office of Research and Development. It is one of the world's largest databases of health and genomic information. MVP partners with veterans receiving care in VA to study how genes affect health. As of November 2019, MVP had enrolled more than 800,000 veterans.

"MVP has enormous potential for increasing our knowledge about the genetics underlying a huge range of traits, including psychiatric traits. It is one of the best samples in the world for this purpose," said Dr. Joel Gelernter, also of the VA Connecticut Healthcare Center and Yale University. Gelernter is one of the senior authors of the work, together with Dr. Murray Stein of the VA San Diego Healthcare System and University of California San Diego.