System Bits: Dec. 26

Adding learning to computer vision
UCLA’s Samueli School of Engineering and Stanford University are working on advanced computer vision technology, using artificial intelligence to help vision systems learn to identify faces, objects and other things on their own, without training by humans. The research team breaks up images into chunks they call “viewlets,” then they have the computer learn how to fit the viewlets into a recognizable object, and next look at other objects to provide some context. “Fortunately, the Internet provides two things that help a brain-inspired computer vision system learn the same way humans do,” says Vwani Roychowdhury, a UCLA professor of electrical and computer engineering and the study’s principal investigator. “One is a wealth of images and videos that depict the same types of objects. The second is that these objects are shown from many perspectives–obscured, bird’s eye, up-close–and they are placed in different kinds of environments.” The researchers turned to cognitive psychology and neuroscience for insights to develop the framework. “Starting as infants, we learn what something is because we see many examples of it, in many contexts,” Roychowdhury says. “That contextual learning is a key feature of our brains, and it helps us build robust models of objects that are part of an integrated worldview where everything is functionally connected.” The researchers tested the system with about 9,000 images, each showing people and other objects. The platform was able to build a detailed model of the human body without external guidance and without the images being labeled. The engineers ran similar tests using images of motorcycles, cars and airplanes. In all cases, their system performed better or at least as well as traditional computer vision systems that have been developed with many years of training.

A new hydrogen fuel catalyst material
Researchers at the University of California, Berkeley, are aiming to replace platinum, an expensive material, as the catalyst for generating hydrogen fuel from water. They resorted to an ingredient originally discovered in jellyfish. No, they actually composed a catalyst with nanometer-thick sheets of metal carbide using a self-assembly process with gelatin, the material that makes Jell-O jiggle. “Platinum is expensive, so it would be desirable to find other alternative materials to replace it,” says senior author Liwei Lin, professor of mechanical engineering at UC Berkeley. “We are actually using something similar to the Jell-O that you can eat as the foundation, and mixing it with some of the abundant earth elements to create an inexpensive new material for important catalytic reactions.” To create the catalyst, the researchers followed a recipe about as simple as making Jell-O. They mixed gelatin and a metal ion — molybdenum, tungsten or cobalt — with water, and then let the mixture dry. “We believe that as gelatin dries, it self-assembles layer by layer,” Lin says. “The metal ion is carried by the gelatin, so when the gelatin self-assembles, your metal ion is also arranged into these flat layers, and these flat sheets are what give Jell-O its characteristic mirror-like surface.” The scientists then heat the mixture to 600 degrees C, which causes the metal ion to react with the carbon atoms in the gelatin, creating large, nanometer-thin sheets of metal carbide, while the unreacted gelatin burns away. “We found that the performance is very close to the best catalyst made of platinum and carbon, which is the gold standard in this area,” Lin notes. “This means that we can replace the very expensive platinum with our material, which is made in a very scalable manufacturing process.”

New vaccine could boost honeybee health
U.S. beekeepers lost an estimated 40% of their honeybee colonies from April 2017 to April 2018 for a variety of reasons, including parasites, pathogens, and pesticides. Researchers are working on a vaccine to deal with the pathogens that can afflict bees. The new PrimeBEE vaccine is administered through a sugar solution meant to attract honeybees while guarding against microbial infections. A queen bee is first vaccinated, then sent to a beekeeper to introduce her to an active hive. The queen then provides immunity to the worker bees in the hive. “Vaccinations have always been more associated with vertebrates and being dependent on the presence of antibodies,” says Dr. Dalial Freitak, now an associate professor in honeybee research at Karl-Franzens University of Graz in Austria. “That’s why vaccinations have never been considered.” The scientists are targeting American foulbrood, a bacterial disease that can easily kill an entire hive once it takes hold among the honeybees. While a vaccine can’t rid honeybees of the mites that can afflict them, it could stave off the viruses that mites carry. Keith Delaplane, director of the Honey Bee Program at the University of Georgia, says, “If an oral vaccine for deformed-wing could be combined with effective mite controls, that would be, in my opinion, a huge leap forward for honeybee health.”