The other day when I picked my three year old son up from daycare, I noticed his teacher had taped an image from a kid’s magazine up on the wall. Beneath it were all of these cute little musings and observations made by the daycare students. “This picture shows a boy and a girl. The boy is sitting on a ball. The girl is being silly.”
When I got to my son’s write-up, it was nearly a page long and he had something to say about everything.
“I think the girl is laughing because the boy has shark teeth on the bottom of his shoes.” (He did!) “The girl has curly hair and the boy likes that her shoes are pink. He likes playing ball games and wants her to be his friend. His skin is a little bit lighter than hers and he has a sweater on…”
The son of a scientist and a writer; he is both observant and prolific.
Devi Parikh, an assistant professor in the Department of Electrical and Computer Engineering at Virginia Tech, has recently won a Google Faculty Award to take a very similar idea and use it to teach computers a little bit of common sense. Although, instead of using toddlers and magazines to generate data, she is using Amazon Mechanical Turk Workers (“Turkers”) and clip art. Through this visual-outsourcing, hundreds of thousands of vetted users from all over the world can create abstract scenes to teach computers object recognition, properties, relationships and poses. These extremely rich scenes touch on something that has never been done before — turning semantic visual features into language.
We all know that (most) kids are afraid of bears. A scene depicting a scared child next to a large, snarling bear would surely make sense to us, but to a computer this is a real head-scratcher. Parikh hopes to use these user-created clip art scenes and depictions to teach computers much of the information we as humans take for granted. In doing so, she will advance computer vision and enhance a variety of machine learning applications, such as autonomous driving.
And as a mother of two young children and oftentimes distracted minivan driver, anything that brings autonomous driving closer to reality definitely gets the green light.
The human mind can only hold so much information at a given time.
Most of us have played the memory game where the host brings out a tray of assorted knick-knacks and gives us a few moments to stare at it before it is whisked away and we are asked to recall as many items as we can.
Generally speaking, this number falls within the “seven plus or minus two” range. One of the most highly cited papers in psychology, “The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information,” speaks to this phenomenon.
Recent advances in Big Data have great promise for solving some of society’s most challenging problems, but how does the human mind handle staring at over a thousand different data points?
Big Data Analytics methods typically require the expertise of a highly skilled analyst. The “human in the loop” method of analysis allows the user to change the outcome of an event or process by modeling different scenarios. Analysts sort through large, complex collections of data sets by employing different models and running simulations. However, most researchers do not possess this level of analytics expertise, and although companies are investing large sums of money to gain insight from the surplus of publically streaming information, they often do not have employees on staff who are able to make much sense of it.
A recent grant award from the National Science Foundation aims to change all of that. Chris North, a professor of computer science and associate director of the Institute for Critical Technology and Applied Science’s Discovery Analytics Center and his well-seasoned team of researchers have won a $1 million grant award to make Big Data Analytics more user-friendly.
The team is using a spatial metaphor system, known as Andromeda, to bring large data clouds down to manageable working sets.
In this system, similar objects are placed in closer proximity. When the user changes the layout of the items, the system updates and learns which data features make those items similar. When people re-organize the data, the data responds and the user is able to learn what features are responsible for that change.
Think back to the aforementioned memory game. Imagine how many more items you could remember if you were able to manipulate the tray and group like objects together to create patterns of similarity. Andromeda allows users to visually interact with the data and glean more meaning from the data points. And using it doesn’t require a Ph.D.
North is excited about the fact that people can apply their own knowledge on the subject area and recognize interesting, new patterns when they arise. The visualization is then tailored to how a person thinks about the data, and the tasks of organization and discovery can occur simultaneously.
By making Big Data a “meaningful picture that users can manipulate,” North and his team are bringing Big Data Analytics to the masses.
Written by Emily Kathleen Alberts
Today kicks off the 2nd Annual Institute for Critical Technology and Applied Science (ICTAS) Program Review.
Join us at the Holtzman Assembly Hall near the Inn with opening remarks by Director Roop Mahajan, followed by Presidential remarks by Timothy D. Sands.
Highly diverse research topics including nanoscale science and engineering, the nano-bio interface, sustainable energy, renewable materials, sustainable water, cognitive systems, and national security are being explored in this two-day event (October 2nd-3rd).
Come find out about our latest research endeavors and hear how we are addressing society’s most challenging needs by transcending traditional academic boundaries.
ICTAS is a premier institute with a mission to advance transformative, interdisciplinary research for a sustainable future.
We hope to see you there!
Ebola virus is headlining across the world, including at the Virginia Tech Carilion Research Institute. Erica Ollmann Saphire will present “The Molecular Toolkit of Viral Hemorrhagic Fevers” as a part of the institute’s 2014–15 Frontiers in Biomedical Research Seminar Series. The lecture will take place on September 12, from 11 a.m. to noon, in R3012 at the Virginia Tech Carilion Research Institute in Roanoke.
A professor in the Department of Immunology and Microbial Science at the Scripps Institute in San Diego, California, Ollmann Saphire studies viruses with compact genomes. Those viruses, including Ebola, offer the most functional “bang” for the polypeptide “buck,” according to Ollmann Saphire. These viruses are coded with only a few proteins, each of which is critically important to the function of the virus.
Ebola, specifically, has seven genes in its genome. Ollmann Saphire and her research team discovered the virus changes function by rearranging those seven genes throughout its lifetime. Ebola looks structurally different as an independent structure than it does as it invades a host. Learning how the physical, molecular changes of Ebola affect the virus’s function has provided invaluable insights for vaccine development.
The Frontiers in Biomedical Research Seminar Series is one of three programs at the Virginia Tech Carilion Research Institute, all of which aim to bring the top scientists to Roanoke. Information on all three programs may be found on the Virginia Tech Carilion Research Institute website.
Science starts with a question. The research is not flashy, even with all of those glinting test tubes and fancy microscopes. It’s slow and specific. Answering that question takes years – sometimes even decades – and that’s just to gather information about one gene or one specific part of a mechanism that might be the solution. There’s no guarantee that the question will ever be answered.
The question is usually big: What causes cancer? Why does this gene mutate? When do neurons age? The path to a solution is usually narrow; it has to be, so how does any one ever choose what to focus on?
When rising fourth-year Virginia Tech Carilion School of Medicine student James Dittmar had to decide on a research project, he was overwhelmed with having to pick just one thing.
“How do you focus?” Dittmar asked. “How do you prioritize research?”
Dittmar explored a number of options with his mentor, Gregorio Valdez, an assistant professor at the Virginia Tech Carilion Research Institute, but none seemed quite right. It was in that exploration that Valdez was inspired to guide Dittmar into finding his ultimate project – helping others decide how to focus their own research. Thus, EvoCor was born.
EvoCor is a free search engine for genes. Type in a gene and EvoCor searches through thousands of mapped genes, different genomes, and larger datasets maintained by the National Institutes of Health. It pulls together a list of genes that evolved similarly. The genes are ranked by likelihood that they’re related functionally to the initial gene submitted.
Take a gene that is already well studied for a certain disease, like MUSK’s role in motor impairment in aging individuals. A scientist can type MUSK into EvoCor and EvoCor will return a list of possibly related genes that might work with MUSK to impair motor function as people grow older.
It’s not a slam-dunk, but it’s a far cry better than picking a random gene that may or may not be related at all. It’s a starting point.
Read the full story here.
What’s in a name? According to Shakespeare, not much. The bard’s well known lines from Romeo and Juliet answer the preceding question thusly: A rose by any other name would smell as sweet.
And if Boris Vinatzer had lived in Shakespeare’s time he would have been able to answer that age-old question with a genome sequence.
But more than sweet-smelling roses, Vinatzer is concerned with pathogens that pose a public health risk. Deadly pathogens like anthrax.
And it’s the example he used in a recently published paper in the journal PLoS One to address limitations with the classic, 200-year-old Linnean system of categorizing organisms.
In his paper Vinatzer proposed using a genome-based naming system.
Because the godfather of genus himself did not foresee the advent of DNA sequencing, a process that reveals genetic similarities and differences at the molecular level the Linnaean system can’t.
The Linnaean system is based on phenotype, or physical appearance, the ease at which names can shift according to discoveries of other like-minded organisms, even after biological life forms have been deputized by scientific organizations, is indeed all too slippery, and that can be frustrating for scientists and a potential public health debacle for government authorities.
The Ames strain of anthrax is one strain of 1,200 that exist in the world. It was dubbed the Ames strain for no other reason than researchers mistakenly thought the sample originally given to the United States Army Medical Research Institute of Infectious Diseases came from Ames, Iowa. Arbitrary? Yes. Effective in communicating vital information about the organism? No.
This particular strain alone is evidence that genome sequencing has relevance beyond a narrow research audience. It gained infamy as the strain that made its way to locations across the United States in the wake of 9/11. Authorities took several months to identify the Ames strain as such. Had there been a bank of genetic samples to run the strain through, the process would have taken days, not months.
What’s in a name? According to Vinatzer and Shakespeare, sometimes not much, unless you peel back the layers to the very molecular bones of an organism.
What if technology could predict traffic patterns the same way meteorologists predict the weather? That is a question graduate student, Hao Chen, has answered through his research under mentor Hesham Rakha, Director of the Center for Sustainable Mobility within the Virginia Tech Transportation Institute and College of Engineering.
At the 2013 Intelligent Transportation Systems World Congress in Japan, Chen won the Best Scientific Paper Award for this research. In his paper, he was not only able to show that this is possible, but also developed a very accurate algorithm for practical applications.
To develop this algorithm Chen used temporal and spatial information to match analogous traffic patterns with real-time and historical traffic data.
Dr. Rakha indicated, “the research that Hao conducted is cutting edge because it not only predicts what will happen on average, but also, the likelihood of certain events. For example the likelihood that your trip will be longer today.” In addition, Dr. Rakha mentioned that “another unique aspect of Hao’s research is the forecast period. Specifically, Hao is predicting traffic conditions up to four hours into the future, something that is far longer than what is reported in the literature (less than an hour). Clearly the longer you predict into the future the more complicated the problem is.”
With the advancement of technology among vehicles and infrastructure, the ability to collect data associated with real time vehicular behavior is becoming a reality. Connected vehicle technology will allow vehicles and roadways to communicate instantaneous information with each other; for example information about vehicle speed, vehicle location, number of vehicles on roadways, changing roadway conditions (such as ice or fog) etc. can become connected or shared.
Therefore, once data from present conditions and historical traffic patterns are measured, the implementation of Chen’s algorithm can predict traffic the same way weather is predicted. Much like a cloud’s anticipated route can be seen through radar, traffic patterns of the past can inform future ones. They can even be represented through a colorful map, similar to the radar map on weather forecast.
This historical traffic data set of roadways in the United States is stored by a company called INRIX. Chen utilized this information to test the accuracy of his algorithm. He looked back at high-traffic scenarios along interstates I-64 and I-264 in Virginia from June and July of 2010 to predict the travel times of the same route on August of 2010. The information of the predicted travel time was updated every five minutes and involved various traffic scenarios. Using his algorithm Chen was able to predict travel time with 96 percent accuracy two to four hours in the future.
Current systems used by most traffic management centers employ only historical data, making them less accurate. While, Chen’s model has the potential to improve accuracy, because it adapts as the environment adapts; as time lapses it increases the amount of historical data collected continually improving. For Chen’s study only two months of traffic data were used, in the future, a growing data set could further narrow the accuracy gap from 96 to 100 percent.
Chen’s algorithm may one day be used to assist drivers in planning their day. This technology has the potential to save drivers a lot of time and money.
The Virginia Tech Transportation Institute conducts research to save lives, time, money, and protect the environment. One of the seven university level research institutes created by Virginia Tech to answer national challenges, the Virginia Tech Transportation Institute is continually advancing transportation through innovation and has impacted public policy on the national and international level.
Parents: if the 1970s cult film Willy Wonka and the Chocolate Factory wasn’t inspiration enough for controlling children’s sugar intake, a recent study touted by the American Heart Association is.
The study, which originally appeared in the journal JAMA Internal Medicine, linked added sugars—such as sugar sweetened beverages, grain-based desserts, fruit drinks, candy, ready-to-eat cereals and more—to a higher likelihood of death by heart disease.
Virginia Tech researcher Brenda Davy has long looked at the effects of high-sugar diets, especially the consumption of sugar sweetened beverages. For a new study, she will test adolescents’ ability to report how much sugar they consume by comparing dietary questionnaire responses with a novel biomarker of dietary sugar intake, delta 13C in the blood.
“It’s often hard to accurately measure the amount and types of sugar that study participants consume, especially children, because we rely on their memory and their ability to accurately estimate quantities,” Davy said. “In nutrition research, we are aware that this is a major challenge when studying the health effects of dietary intake in children. Biomarkers can provide more accurate and objective assessments of dietary habits, so we hope to receive additional funding from NIH to support this project. Another exciting aspect of this project is that it involves collaboration with a faculty member in Hawaii, Dr. Hope Jahren, who is an expert in geochemistry and geophysics”.
Davy and graduate student Shaun Riebl, a registered dietitian and Ph.D. candidate in the department of human nutrition, foods, and exercise, are looking for 75-100 local participants between the ages of 12 and 18 to participate in a 2-3 week long study. During this time, respondents will visit Virginia Tech’s campus 4 or 5 times; two visits will involve providing a routine fingerstick blood sample.
The goal is to have 100 respondents by the end of July, according to Riebl.
“Our work does not seek to make sugar another one of the ‘bad guys’ like many nutrients have become. A problem can come about when too much sugar is eaten,” Riebl said. “Our hopes are to provide another piece to the complicated nutrition puzzle helping concerned parents identify potential unhealthy habits in their child’s diet choices so changes can be made before a visit to a doctor is necessary. We want children and their parents to feel good and be healthy today and in years to come.”
It’s not just about mooo-ving herds, or perfected grain.
The contemporary animal science field charges beyond traditional animal production tactics to focus on biologically-driven animal health research, according to Dave Gerrard, head of Virginia Tech’s Animal and Poultry Sciences department.
Studying animal health—such as muscle biology, genetics, reproduction, nutrition, and physiology—contributes to better animal well-being, but also provides an excellent foundation for studying human health, Gerrard says. The evolution of the field is in line with Virginia Tech’s push to become an academic front runner in the health sciences.
A team of animal science researchers focusing on health and disease were purposefully hired to the Animal and Poultry Sciences department in the last two years, according to Gerrard. Placental biologist Alan Ealy studies how various physiologic, metabolic and environmental stresses during early pregnancy impacts reproduction in cows and sheep, and uses it as a model for understanding similar processes in humans.
“Embryonic and fetal development is remarkably similar between cattle, sheep and humans, and exploring various developmental processes and events will provide us with clues into how we may optimize neonatal and lifetime health in humans and efficiency of production in livestock,” Ealy said.
Meanwhile, new hire Shelly Rhoads studies how adolescent obesity affects pigs—particularly during reproduction— with the expectation that it will provide some insight into human childhood obesity. In other words, if one gets fat before puberty, how will that affect reproductive success?
“Pig and human physiologies are very similar,” said Rhoads. “A pig is more similar to humans than traditional lab mice.”
Early results in the pig model show that what a pregnant mother eats, rather than how much she weighs, has a greater impact on reproductive success. Diets composed of a high amount of the sugars fructose and glucose (typically found in soft drinks) have a greater impact than fatty tissue, Rhoads said.
Other research topics among the new hires include the use of stem cells to fix suspensory ligaments in race horses and the neurobiology of appetite in chickens.
“We are using top-end science to solve problems pertinent to agriculture and human medicine,” Gerrard said. “I’m looking forward to seeing where these cutting-edge approaches take us.