Duke Research Blog

Following the people and events that make up the research community at Duke.

Author: Olivia Zhu (Page 1 of 5)

The Art of Asking Questions at DataFest 2016

During DataFest, students engaged in intense collaboration. Image courtesy of Rita Lo.

Students engaged in intense collaboration during DataFest 2016, a stats and data analysis competition held from April 1-3 at Duke. Image courtesy of Rita Lo.

On Saturday night, while most students were fast asleep or out partying, Duke junior Callie Mao stayed up until the early hours of the morning pushing and pulling a real-world data set to see what she could make of it — for fun. Callie and her team had planned for months in advance to take part in DataFest 2016, a statistical analysis competition that occurred from April 1 to April 3.

A total of 277 students, hailing from schools as disparate as Duke, UNC Chapel Hill, NCSU, Meredith College, and even one high school, the North Carolina School of Science and Mathematics, gathered in the Edge to extract insight from a mystery data set. The camaraderie was palpable, as students animatedly sketched out their ideas on whiteboard walls and chatted while devouring mountains of free food.

Callie Mao ponders which aspects of data to include in her analysis.

Duke junior Callie Mao ponders which aspects of the data to include in her analysis.

Callie observed that the challenges the students faced at DataFest were extremely unique: “The most difficult part of DataFest is coming up with an idea. In class, we get specific problems, but at DataFest, we are thrown a massive data set and must figure out what to do with it. We originally came up with a lot of ideas, but the data set just didn’t have enough information to fully visualize though.”

At the core, Callie and her team, instead of answering questions posed in class, had to come up with innovative and insightful questions to pose themselves. With virtually no guidance, the team chose which aspects of the data to include and which to exclude.

Another principal consideration across all categories was which tools to use to quickly and clearly represent the data. Callie and her team used R to parse the relevant data, converted their desired data into JSON files, and used D3, a Javascript library, to code graphics to visualize the data. Other groups, however, used Tableau, a drag and drop interface that provided an expedited method for creating beautiful graphics.

Mentors assisted participants with formulating insights and presenting their results

Mentors assisted participants with formulating insights and presenting their results. Image courtesy of Rita Lo.

On Sunday afternoon, students presented their findings to their attentive peers and to a panel of judges, comprised of industry professionals, statistics professors from various universities, and representatives from Data and Visualization Services at Duke Libraries. Judges commended projects based on aspects such as incorporation of other data sources, like Google Adwords, comprehensibility of the data presentation, and the applicability of findings in a real industry setting.

Students competed in four categories:  best use of outside data, best data insight, best visualization, and best recommendation. The Baeesians, pictured below, took first place in best outside data, the SuperANOVA team won best data insight, the Standard Normal team won best visualization, and the Sample Solution team won best recommendation. The winning presentations will be available to view by May 2 at http://www2.stat.duke.edu/datafest/.

Bayesian, the winner of the Best Outside Data category

The Baeasians, winner of the Best Outside Data category at DataFest 2016: Rahul Harikrishnan, Peter Shi, Qian Wang, Abhishek Upadhyaya. (Not pictured Justin Wang) Image courtesy of Rita Lo.

 

By student writer Olivia Zhu  professionalpicture

“Debugging the Gender Gap” in Tech

Lenna“Why isn’t Lenna wearing any clothes?” I implored my friend, shocked at seeing the shoulders-up nude photo of a woman on a mundane Monday in the Duke library. I had been going through a MATLAB tutorial on computer vision, and the sample image was, surprisingly, a naked lady. Apparently, when the USC developers behind a computer vision algorithm needed a sample face in 1973, someone just happened to walk into the lab with a Playboy magazine. The face of the woman on the centerfold, Lenna, has since become the default data for computer vision classes around the world. Because, of course, it’s totally normal to walk into an academic setting waving around a copy of Playboy, which would naturally be the first place one would go looking for a face.

Unfortunately, seeing female objectification in professional programming environments isn’t exactly an isolated incident. With the advent of the “brogrammer” culture, women have reported being exposed to workplaces in which male programmers share porn over open communication channels, according to CODE: Debugging the Gender Gap. When they’ve asked their male coworkers to stop, they were told, “Stop being such a girl.”

A showing of CODE was put on by RENCI, the Renaissance Computing Institute, and the

new doc 6_1 (1)

The percentage of women earning degrees in computer science has been decreasing, rather than increasing, since the 1980s.

Carolina Women’s Center, on February 29 at UNC. RENCI, while addressing issues of staffing diversity within its own organization, was inspired to bring the issue to light in the greater UNC community. By 2020, we expect to see more than one million unfilled software engineering jobs. As of now, only 23% of technical jobs nationwide are filled by women, leaving a huge gap to fill in this important workspace.

The response of the largely female audience to the film was overwhelmingly positive. Lilly, a first-year math student at UNC, noted that the issues the film addressed were “obvious,” both in academic settings and in the online blogosphere. She appreciated the positive messages, such as in this GoldieBlox superbowl ad, that counter expectations of young girls to study more “social” subjects and encourage them to pursue science, technology, engineering and math. Addy, a first-year computer science student, noted that a supportive group of women in her CS401 class at UNC makes the dearth of women less noticeable.

Olivia, Tabatha, and Addy with a collage of "Why We Love Tech"

Olivia, Tabatha, and Megan with a collage of “Why We Love Tech”

Tabatha, a first-year computer science student at UNC, said that she feels intimidated in introductory computer science classes, where male students often have years of background knowledge that she doesn’t. She hesitates to show men her code until it is perfect, since she feels that as a woman, she has to prove that she is just as good as a man. This additional pressure and worry, CODE observed, often causes women to perform worse in quantitative classes. Tabatha, Megan, and Olivia attended the screening as part of a Women’s Studies class. Megan echoed Tabatha’s sentiment, relating that as a beginning programmer, she felt behind during HackNC, where most men already knew how to build apps.

Clearly, issues of female representation in tech persist into the university and industry level. However, CODE insists that we must remedy the problem during childhood, when girls receive societal messages that deter them from studying science and tech subjects.

If we’re going to be “changing/saving the world,” “making a better version of you,” and deciding how to “do the right thing,” (all rhetoric from the tech industry), we should probably have all genders and races represented in those responsible for effecting the change that will supposedly impact all of humanity.

For more information on CODE, check out shescoding.org.

By Olivia Zhuprofessionalpicture

An Adventure Abroad in Brain-Machine Interfaces

11080630_10205422939006642_2749326952690554776_o copyMatthew McCann, Pratt ’16, spent his summer translating thoughts into movements.

A biomedical engineering and mathematics major, the Duke senior contributed to work in the field of prosthetics by creating a brain-machine interface that senses different brain waves of a subject and converts them into movements of a mechanical hand.

McCann, who had never traveled to Europe, let alone lived there for three months, took his foreign adventure one step further and pursued cutting-edge research in Rand Almajidy’s biomedical engineering lab in Germany. McCann was paired with the University of Freiburg for a Research Internship in Science and Engineering by the German Academic Exchange Service.

McCann combined two prominent biomedical techniques, tri-polar concentric electroencephalograms (tEEG) and near-infrared spectroscopy (NIRS), to pick up the brain activity of his subjects. EEGs are the typical devices one pictures when imagining recording brain activity: electrodes stuck all over a subject’s skull to pick up neuron firing when particular brain regions are active.

NIRS is a novel way of measuring brain activity. A common application of NIRS is in the pulse oximeter, or the plastic clip-like contraption doctors place on your finger to measure pulse and blood oxygenation. McCann used NIRS to measure the blood flow in different regions of the subject’s scalp. Different patterns of blood flow indicated dynamic brain activity.

Based on data obtained from these two techniques, McCann categorized brain activity into three specific intentions: thinking about moving the right hand, thinking about moving the left hand, and thinking about moving the feet. Each different intention to move was then connected with moving one finger of a mechanical hand. An example of the hand moving in response to different intentions is shown below (at 8x speed):

McCann’s major challenges in the project were processing complicated EEG signals and removing noise from these signals in order to correctly classify each of the movement intentions. He worked with vast amounts of training data from subjects who had practiced focusing acutely on each of the movements.

He ultimately isolated the specific frequency bands whose power was modulated most drastically during the three movement intentions he was targeting. These frequency bands served as the basis for his machine-learning algorithm, which matched known data the subjects had been trained to produce with unknown thoughts about movement.

After developing his algorithm, McCann tested it on unknown data, in which subjects thought about moving their right hand, their left hand, and their feet in some arbitrary sequence. McCann’s algorithm ultimately obtained impressive accuracy of up to 80% when categorizing unknown thoughts about movements.

Through his research, McCann demonstrated the feasibility of rapidly creating functional prosthetics from simple materials and only open-source software. His prosthetic hand proves promising to medical innovation, as it represents a non-invasive, functional brain-machine interface. Ultimately, his success sheds optimism on the future of prosthetics.

Learn more about McCann and his projects on his website.

professionalpictureby Olivia Zhu

Making Sense of Noise: Stephen Lisberger

Imagine catching a ball thrown at you out of mid-air. Your response seems almost instinctive, like a reflex. However, this seemingly simple movement contains complex components: one must judge the ball’s arc to decide where it will intersect a particular height, and how fast one must move his hand to catch the accelerating ball.

This calculation requires an entire concert of neural signals, firing in a manner so precise that it produces an accurate estimate of the speed and direction of the ball’s trajectory. Add to this complicated model the fact that each individual neuron produces a certain amount of noise — that is, across various trials, the same neurons produce different firing responses to the same stimuli. These multiple layers of convolution would frustrate most, but Dr. Stephen Lisberger thrives upon it.

Screen Shot 2016-01-27 at 12.47.43 PMLisberger, the Chair of Neurobiology at Duke School of Medicine, emphasizes that while a single noisy neuron cannot produce an accurate estimate of speed and direction, the key lies in populations of neurons. On January 25, Lisberger presented his research to a diverse crowd of Duke scientists.

Lisberger and his team have performed multiple trials in which a monkey tracked a visual stimulus with his eye, thus activating certain neurons. They found that the noise persisted even in neural populations.

Lisberger, rather than being discouraged, turns this noise into an asset. He reasons that variation is something which the brain must handle; therefore, he can use variation to learn about the brain.

When a monkey follows a visual stimulus with his eye, he integrates the sensory system with a motor region of the brain called MT. Lisberger isolated the source of the noise to the sensory system, rather than MT. He found that other movements originating from MT did not display the same noise; thus, the noise in eye tracking must have come from the sensory system.

The noise from the sensory system propagates down to MT, and Lisberger follows in his analysis.

One of his colleagues proposed that the random noise over a large population of neurons should cancel itself out. Lisberger contradicts this idea, noting that the variation is correlated among neurons in MT. Variations in pairs of neurons fluctuate up and down together. Thus, some of the “noise” is actually signal. This shared noise is transmitted through the circuit, while independent noise averages itself away.

Ultimately, Lisberger models neural responses over multiple trials to statistically estimate the direction and speed indicated by a particular response. The brain though, has not the luxury of simultaneously integrating and analyzing such large pools of data in its fraction-of-a-second estimate. Instead, the brain makes do with what it has, which, as Lisberger points out, is enough.

By Olivia Zhu  professionalpicture

HTC Vive: A New Dimension of Creativity

“I just threw eggs at the robot!” grad student Keaton Armentrout said to Amitha Gade, a fellow biomedical engineering master’s student.

image2

“He just said, ‘Thank you for the egg, human. Give me another one.’ It was really fun.”

In what world does one throw eggs at grateful robots? In the virtual world of the HTC Vive, a 360 degree room-size virtual reality experience created by Steam and HTC that is now offering demos on the Duke campus from November 9 – 13. There is a noticeable buzz about Vive throughout campus.

I stepped in to the atrium of Fitzpatrick CIEMAS expecting a straightforward demonstration of how to pick up objects and look around in virtual reality. Instead, I found myself standing on the bow of a realistic ship, face to face with a full-size blue whale.

A Tiltbrush drawing I created with HTC Vive during my internship at Google. (Tiltbrush was acquired by Google/Alphabet).

A Tiltbrush drawing I created with HTC Vive during my internship at Google. (Tiltbrush was acquired by Google/Alphabet).

Peering over the side of the shipwreck into a deep ravine, I seriously pondered what would happen if I jumped over the railing –even though both my feet were planted firmly on the ground of CIEMAS.

Armentrout observed that the Vive differentiates itself from other VR devices like Oculus by allowing a full range of motion of the head: “I could actually bend down and look at the floorboards of the ship.”

In Valve’s Aperture Science demo, based on their game Portal, I attempted to repair a broken robot so real it was terrifying. I was nearly blown to bits by my robot overseer when I failed at my task. In total, I progressed through four modules, including the shipwreck, robot repair, a cooking lesson, and Tiltbrush, a three-dimensional drawing experience.

Game developers naturally are pursuing in virtual reality, but technologies like HTC Vive have implications far beyond the gaming realm. One of the applications of the Vive, explained one of the Vive representatives, could be virtual surgeries in medical schools. Medical schools could conserve cadavers by assigning medical students to learn operations on virtual bodies instead of human bodies. The virtual bodies would ideally provide the same experience as the operating room itself, revolutionizing the teaching of hands-on surgical skills.

Gade brainstormed further potential applications, such as using robots controlled by virtual reality to navigate search-and-rescue situations after a crisis, reducing danger to rescue crews.

The first time I tried the HTC Vive was not at Duke; it was at a Tiltbrush art show in San Francisco.

HTC Vive Tiltbrush masterpiece displayed at the San Francisco Tiltbrush art show

HTC Vive Tiltbrush masterpiece displayed at the San Francisco Tiltbrush art show

On the stage, an artist was moving her limbs in grand arcs as she painted the leaves of trees and brushing the ground to create a sparkling river. A large screen projected her virtual 3-D masterpiece for the audience.

Gilded frames on stands emphasized the interactive Vive devices, each of which housed a Tiltbrush masterpiece created by a local artist trained in the technique. Well-dressed attendees marvelled at seemingly invisible waterfalls and starry skies in the virtual reality paintings. Clearly, the Vive, by opening another dimension of artistic creation, is changing our notions of space and pushing the bounds of creativity.

12188016_10204922617616904_5669989382191630573_oBy Olivia Zhu Olivia_Zhu_100

From Neutrinos to Nuclear Deals: Congressman Bill Foster

Hon. Bill Foster of the 11th District of Illinois is the only member of Congress to hold a Ph.D. in science. On November 5th, Congressman Foster visited Duke’s Initiative for Science and Society to discuss his unconventional path to politics and his consequent unique perspective. He lightheartedly delivered what he called a “recruiting speech” to a room full of scientists, hoping to persuade students with scientific background to become involved in public policy.

Representative Bill Foster, Ph.D., doing what politicians must.

Bill Foster started his first business with his brother at the age of 19 out of his family basement. His earnest, innovative efforts to use computers to control lighting manifested in the company Electronic Theatre Controls, which powered Disneyland and Disneyworld’s Parade of Lights in the 1980s, the 2012 London Olympic Stadium, Chicago’s Millenium Park, and a large portion of shows on Broadway.

Foster then transitioned into his career in physics. He undertook the IMB Proton Decay Experiment for his Ph.D. thesis under Larry Sulak; Foster did not observe proton decay, but he did observe neutrinos from a supernova. Foster continued his physics career at the Fermi National Accelerator Lab in suburban Chicago, where he smashed protons and anti-protons together at high speeds and later worked on the particle accelerators themselves.

In the midst of discovering Big Bang particles, Foster also fell into politics by maintaining an active civil engagement. He volunteered for Patrick Murphy’s campaign in 2006, where he says he “learned business on the factory floor,” a philosophy he has maintained since his days at Electronic Theatre Controls. He began the 110th Congress as an intern for Rep. Patrick Murphy, and ended it sitting as a Congressman.

Hon. Foster graphs the relative numbers of scientists and engineers, lawyers, and career politicians in Congress. The U.S. Congress consists mostly of career politicians, explains Foster, while China, for example, consists mostly of engineers.

Rep. Foster plots the relative numbers of scientists and engineers, lawyers, and career politicians in international governing bodies. The U.S. Congress consists mostly of career politicians, explains Foster, while China, for example, consists mostly of engineers.

Since winning his seat in 2012, Foster has introduced a scientific perspective to Congress, even if he’s careful not to conflate that with his political stance. He makes a point to clarify technical details of issues like the Iran nuclear deals, human genetic engineering, and public key cryptography on cell phones, to ensure that Congress makes the most informed decisions possible on highly complicated ethical issues. On genetic engineering, he noted, “Our ethical paradigm is not set up for it,” as the notion of “All men are created equal” fundamentally cannot handle humans whose genetic traits are pre-picked. Clearly, scientific expertise will be invaluable in such consequential issues.

Life in Washington, Foster stated, is unromantic. Foster lives in efficiency apartments and grounds himself by holding “Congress on your Corner” events, where he answers any constituent questions, like why grout isn’t working on a driveway.

Political customs, such as the dilemma of which tie to wear to promote his campaign, still bewilder his scientific mind. Most of the votes he makes, like renaming a post office, or voting on an issue the President will inevitably veto, don’t really matter, he said.

But what makes politics worth it for him, Foster explained as he passed around his voting card, is the power to make a positive difference in issues that impact millions of people. Such ambitions transcend the boundaries between science and policy.

By Olivia Zhu Olivia_Zhu_100

Page 1 of 5

Powered by WordPress & Theme by Anders Norén