Skip to Content

CSE News

  • CSE Alumnus, Lead Engineer for Pokémon GO, Back on Campus

    Next time you see someone playing Pokémon GO, the popular mobile phone-based game, keep in mind that a Computer Science and Engineering alumnus, Edward Wu (B.S. '04) leads the groundbreaking game’s technical team.

    Wu is a senior product manager at Niantic, the company that makes Pokémon GO and was spun off in October 2015 from Google, where Wu worked previously.  He earned a dual bachelor’s degree in computer science and physics, and what he learned at UC San Diego is the basis of his success as an engineer, he said during a talk on campus Oct. 13 organized by the Center for Networked Systems (CNS). “I learned the core algorithms, the core fundamentals here,” Wu said. “There is no substitute for that.”

    Wu stayed in touch with CSE Prof. Geoffrey Voelker on and off since 2004, and it was Voelker who invited Wu to speak on campus as part of CNS's Fall 2016 research review. “Ed is an example for all our students to show that what they’re learning prepares you to go out into the world and make a difference,” said Voelker. “The world is now a different place because of Pokémon GO.”

    In his talk, Wu gave an overview of all the engineering and troubleshooting that has to happen for users to catch Pokémon, get supplies, and battle in gyms on their smartphones, at any time and in any place from the United States to France, to Australia. “The key element is overlaying a single, consistent reality over the real world,” Wu explained.

    This is all the more challenging because the game has been downloaded by more than 500 million people. Making Pokémon GO work for even a small fraction of these users is no small feat. Wu and his team spent most of July 2016 in a sleepless state while they were launching the game around the world. Demand was 50 times more than Niantic projected.

    But Pokémon GO is more than just a game, Wu said. “It’s about going outside, going on walks and meeting people in the real world,” he said. The game requires players to walk around and hit up designed spots, called Pokéstops, to get supplies. Players need to physically be near the gym where they want to do battle. Players have logged more than 4.6 billion kilometers (about 2.8 billion miles) between the game’s launch in July and August of this year—that’s half the distance between Earth and Pluto.

    Niantic also recently introduced a feature that allows players to get rewards to power up and evolve Pokémon for every kilometer (about 0.6 miles) they walk with their favorite Pokémon. Wu’s walking buddy is Psyduck, which looks like a cross between a yellow duck and a platypus, walks upright and has psychic powers.

    During his CNS talk, Wu recalled how he tried his hand at developing a multiplayer game for the first time in CSE 125, a computer science class taught by Voelker. Wu and the rest of a student team created a real-time tactical combat game they called “Geteilte Stadt,” German for “a city divided.” During the class, he learned how to collaborate and work with others on complex technical problems, he said. He learned how to code, by himself and with others, and how to resolve disagreements around technical issues. “It was invaluable,” he said. Wu wore a tuxedo during the CSE 125 final presentations, when all teams demonstrated their games.

    Wu was a Jacobs Scholar as an undergraduate at UC San Diego—a select group chosen for their academic achievements, leadership potential and commitment to community service. Jacobs Scholars receive full tuition and living expenses, as well invitations to cultural and other social events hosted by Joan and Irwin Jacobs, and access to a network of current and former Jacobs Scholars. In 2003, he was also a Calit2 Summer Undergraduate Scholar.

  • Dual Poster Sessions Showcase CSE Research at FISP Symposium 2016

    At least 72 posters will be on display when UC San Diego students share their research results during two poster sessions during the Frontiers of Innovation Scholarship Program (FISP) Symposium on Tuesday, October 18 in the Price Center. (Click link at bottom to read a preview of oral presentations at the symposium.)  Hour-long poster sessions are scheduled to start at 12:30pm and 4:45pm, respectively, in Price Center Ballroom A. Posters are only on view during one session not both. (Consult the FISP program for details.)

    The presenters will include CSE fourth-year Ph.D. student Mohsen Malmir (right), who will present "Music Generation by Deep Recurrent Neural Networks."  He was mentored by CSE Prof. Gary Cottrell and Music Prof. Shlomo Dubnov. Malmir developed algorithms that learn to generate new music by learning from annotated music. The student trained "deep recursive neural networks (DRNNs) to generate sequences of tones along with their temporal information," says Malmir, adding that he aimed to "analyze the structure of the trained network and the learned sequences to find relations with higher level structures in music." Subsequent to the project, Malmir proposes to make the music dataset and relevant codes available to the research community for further development.

    CSE computer science junior Allan Yeh (left) will present a poster on "Interacting with Chemical Software." Yeh has been developing a molecular database in which new molecules can be uploaded for users to download and use in their own molecular simulations. Says Yeh: "The FISP scholarship has helped give me a better understanding of how my knowledge of computer science can be applied to the world at large."

    Structural Engineering professor and CSE faculty-affiliate Falko Kuester mentored multiple poster presenters and speakers from different majors, including electrical and computer engineering as well as visual arts.

    ECE senior and Qualcomm Institute intern Dimitri Schreiber (right) will present a poster on "CAVECamX: Autonomous Stereo Spherical Panorama System."  CAVECamX is a small, binocular, two-axis gimbal system used for creating high-resolution 3D photospheres, combined with GPS and inertial measurement unit (IMU) data, enabling better coregistration internally within a single photosphere, and externally between heterogeneous datasets, including fusion with point clouds generated from photogrammetry and LIDAR" laser scanning. "This decreases human processing time by automatically recording location and orientation of the dataset, which would previously be recorded manually and therefore likely left out or lost," says Schreiber. "The attitude data will help fully automatic stitching of stereoscopic datasets without the commonly associated motion sickness by constraining the system." Noting that CAVECamX is small and consumes little energy, the student adds that it "enables remote visualization of archaeological sites, allowing researchers to be virtually immersed in the captured scene without having to travel across the globe."

    Kuester also mentored two Visual Arts majors -- Samuel Balatbat (below left), and Bertha Yue (below right) -- both working on 3D visualization projects involving the Chaco canyon archaeological site in New Mexico.

    Balatbat's project, "Stereoscopic Photospheres of the Historical Site of Chaco in New Mexico," involves stereoscopic photospheres, i.e., pairs of images captured by either a CAVEcam or CAVECamX. "These images have full 360-degree depth coverage, making it possible for archaeologists and viewers to be virtually present at sites," according to Balatbat's abstract. The visualization can happen through either head-mounted displays (such as the Oculus Rift or HTC Vive) or through the Qualcomm Institute's large-scale CAVE display systems (WAVE, StarCAVE, NexCAVE, etc.). Balatbat will showcase some of the stereoscopic photospheres captured at the Chaco site and processed for 3D viewing.

    Bertha Yue's research also focused on the New Mexico site. Her poster will explore "Using Photogrammetry to Create a 3D Model of the Chaco Pueblos in the American Southwest." (Her research abstract was deleted from the FISP Symposium program "owing to proprietary information.")

    Read more about oral presentations on the FISP Symposium agenda.

  • CSE Highlights Research at Frontiers of Innovation Symposium

    The second annual UC San Diego Frontiers of Innovation Scholars Program (FISP) Symposium will take place on October 18 from 8am to 5:45pm at the Price Center. The symposium will showcase interdisciplilnary research carried out by postdocs, graduate and undergraduate student researchers with funding from the campus itself. Two types of presentations are scheduled: oral presentations, and poster sessions (for one-on-one interaction with the student researchers). The oral presentations will run 15 minutes each for undergraduates, 25 minutes for grad students and postdoctoral researchers (in both cases, leaving 5 minutes for Q&A at the end of each talk). 

    CSE students set to deliver oral presentations at the FISP Symposium include Sharad Vikram, a first-year graduate student specializing in machine learning under his mentor, CSE Prof. Sanjoy Dasgupta. His topic: "Air Quality Monitoring with Cheap Hardware."  Vikram designed a cheap air-quality sensor that monitors CO, NO and other pollutants with the goal of better understanding and eventually improving air pollution patterns over a large area (e.g. San Diego County). "We are currently collecting a dataset of sensor measurements from some select locations in Los Angeles," said Vikram. "Current sensors are expensive and immobile, but will produce more reliable and precise measurements than those from a sensor with commodity hardware. Future challenges include remote calibration of sensors to produce robust measurements and inference of air pollution in areas without sensors." Using machine learning and statistical inference, Vikram aims to solve such problems.

    Dual computer science and mathematics major Carolyn Breeze and two other undergraduates in anthropology, Rosemary Elliott Smith and Taylor Harman, worked together on the interdisciplinary At-Risk Cultural Heritage project funded by a University of California Catalyst award. The students were mentored by principal investigator Tom Levy, who directs the new Center for Cyber-Archaeology and Sustainability (CCAS), based in the Qualcomm Institute. The joint FISP project was presented as part of an all-themes session because of its broad applicability to all FISP target areas. The project, "At-Risk Cultural Heritage and Archaeological Data Management: The ArchaeoSTOR Solution," involves a web-based database (above left) called ArchaeoSTOR that helps researchers safely store artifact metadata, location data, photographs, and even point-cloud data (produced using LIDAR laser scans). The students traveled to Greece this past summer with Levy to participate in excavation of a looted Mycenaean tomb at the site of Kastrouli near Delphi. According to their abstract, "the Kastrouli excavations proved to be a perfect field test for the applications of ArchaeoSTOR that our team developed [because it allowed them to] dramatically improve the functionality of the database and preserve vast amounts of precious data associated with the cultural heritage site." 

    Given the inherently interdisciplinary nature of FISP research projects, not all the students working under CSE mentors were computer science students themselves. For instance, CSE Prof. Gary Cottrell mentored nanoengineering Ph.D. student Chen Zhang (right), who was developing "Small Molecule Accurate Recognition Technology" (SMART) while also working in the Center for Marine Biotechnology and Biomedicine in the Scripps Institution of Oceanography. The tool itself was developed to speed up marine natural products discovery, i.e., natural products found in the ocean. "By testing different spectra using this algorithm," says Zhang, "we can rapidly generate hypotheses about the relationship of new molecules to those used for the training -- based entirely on their nuclear magnetic resonance properties."

    CSE professors Ryan Kastner, faculty-affiate Falko Kuester and Scripps Prof. Stuart Sandin together mentored Clinton Edwards, a Scripps grad student. Edwards (left) will present on a "Platform for Ocean Imaging: Building Capacity for Visualizing, Analyzing and Communicating Underwater Ecological Data." The project targets documentation of large plots of seafloor habitats -- measuring hundreds of square meters -- by speeding up the post-processing that is so time-consuming and computationally intensive. "We have begun to develop and test platforms to address the intensive collection, storage and processing steps required to facilitate rapid extraction of key metrics from 3D digital maps of the seafloor," writes Edwards in his abstract. "These maps will enable new insights in community ecology by increasing the scale of observation by over an order-of-magnitude [scale] larger than what is currently available."

    Kuester is also the moderator of an undergraduate panel on enriching human life and society, and he mentored the first speaker of the session: UC San Diego media studies major and newly-minted alumna, Emily Zheng (B.A. '16). Her talk will be on "Media in the Field". Zheng (right) is a media intern in Kuester's CISA3-CHEI (Cultural Heritage Engineering Initiative). She was responsible for producing content based on CISA3-CHEI expeditions. In her presentation, Zheng will focus on media produced on expeditions to San Marino and Chaco Canyon. "As a part of CHEI, my team collects data for the purposes of 3D reconstructions. This is done through techniques such as LIDAR scanning, SFM [Structure from Motion], CaveCams, and UAV imaging," notes Zheng. "My work concerns the recording of the labor behind data collection as it happens in the field to show the techniques and challenges of working in variable and changing situations." (Zheng was also peer-mentored by Dominique Meyer, a previous recipient of a FISP scholarship to work  with Kuester.)  

    "What I found impressive is that our FISP students applied their research at not just one, but two UNESCO World Heritage Sites, conducting fieldwork at home and abroad, in San Marino and Italy," observed CISA3-CHEI director Kuester. "They also presented their work at national and international conferences, and they leveraged opportunities with two other NSF and National Geographic-funded archaeology projects outside of UC San Diego. These undergraduate trainees left quite a legacy over just the past year." 

    CSE lecturer and Qualcomm Institute (QI) research scientist Jurgen Schulze and ECE Prof. Truong Nguyen jointly mentored ECE second-year Ph.D. student Ji Dai on a study of "Stereo Panorama Generation from Point-Cloud Re-projection." Dai (left) will propose an "algorithm that can produce stereo panoramas with minimized vertical disparity and parallax error. The algorithm only needs input stereo image pairs and the camera positions at which the image pairs were captured. The algorithm also provides users the freedom of choosing desired viewing location and angle." They can also choose a preferred baseline.

    For information on CSE involvement in the FISP poster sessions, see separate article.

    Download the complete program PDF with abstracts for all presentations.

  • Computer Vision Papers Presented at ECCV 2016 in Amsterdam

    The European Conference on Computer Vision runs Oct. 8-16 in Amsterdam, and UC San Diego's Center for Visual Computing (VisComp) is heavily represented at the international forum that is among the premier academic gatherings on computer vision. Two CSE professors -- VisComp director Ravi Ramamoorthi and Manmohan Chandraker -- were among the authors of eight VisComp papers presented at ECCV 2016.

    Professor Ramamoorthi was the senior author on "A 4D Light-Field Dataset and CNN Architectures for Material Recognition." The paper was joint with colleagues from UC Berkeley as well as recent CSE faculty arrival Manmohan Chandraker and visiting industrial fellow Ebi Hiroaki from Sony, one of VisComp's founding industry partners.) The paper focused on the use of deep learning for recognizing materials using 4D light-field (LF) photography taken with a Lytro Illum 4D LF digital camera. Pictured at right: In recognizing materials, top grids show 4D light-field predicts more accurately than 2D inputs, while bottom grids show 2D more accurate. Conclusion: light-field recognition performs best when object information is missing or vague, so must rely on local texture, viewpoint change or reflectance information available with 4D light-field imagery.
    "Our main goal [was] to investigate whether the additional information in a light-field (such as multiple sub-aperture views and view-dependent reflectance effects) can aid material recognition," noted the authors, who reported a seven percent boost with the best-performing convolutional neural network (CNN) architecture compared with standard 2D image classification. "Our dataset also enables other novel applications of light-fields, including object detection, image segmentation and view interpolation."
    Another of Ramamoorthi's papers was co-authored by colleagues at the University of York (UK), and Sapienza-University of Rome (Italy). University of York's Will Smith was a sabbatical visitor at UC San Diego from York in Winter 2016 when the research took place. The collaborators presented a method for estimating surface height directly from a single polarization image simply by solving a large, sparse system of linear equations. The paper, "Linear depth estimation from an uncalibrated, monocular polarization image," is available online. 
    In addition to collaborating with Ramamoorthi, CSE Prof. Manmohan Chandraker had a paper at ECCV 2016 on a "Deep Deformation Network for Object Landmark Localization." The work was done while Chandraker was still a research scientist at NEC Laboratories America, before taking up his professorship in CSE earlier this year. His coauthors on the paper were also working in NEC Labs' Department of Media Analytics at the time.
    The Center for Visual Computing is an interdisciplinary research unit, and other faculty with papers at ECCV 2016 included Electrical and Computer Engineering's Nuno Vasconcelos and Cognitive Science professor Zhuowen Tu. Tu has an appointment in CSE as well, so he can supervise CSE Ph.D. students, as he did with first author and CSE Ph.D. student Saining Xie (at left) on a joint paper about "Top-Down Learning for Structured Labeling with Convolutional Pseudoprior."  The authors proposed a new method for structured labeling by developing convolutional pseudo-prior (ConvPP) on the ground-truth labels, and they reported "state-of-the-art results on sequential labeling and image labeling benchmarks."

    Learn more about VisComp papers presented at ECCV 2016 on the Jacobs School blog.  

  • Students, Alumni Take Top Spots at San Diego Zoohackathon

    Over the October 7-9 weekend, a small group of programmers participating in the inaugural Zoohackathon had the opportunity to go behind the scenes at the San Diego Zoo Safari Park and see firsthand some of the species the San Diego Zoo Global has rescued from illegal wildlife trade. “The zookeepers brought out two monitor lizards that were brought into the United States illegally and confiscated,” said Utkrisht Rajkumar, a third year computer engineering major at the University of California San Diego. “It added a lot of context to the problems that we were trying to solve.”

    The group hunkered down for the weekend for a two-day invention session aimed at developing usable solutions to problems solicited from wildlife experts around the world. Leading conservation technology zoos in the U.S., UK, Asia and the Pacific are running similar events.

    At the end of the weekend, teams pitched their ideas to an expert panel of judges. The first-ever San Diego Zoohackathon took place at the Institute for Conservation Research in Escondido, Calif.

    One of those problems presented to the participants is that many illegal wildlife products, such as rosewood from protected forest habitats or ivory figurines made from elephant tusks, are confiscated upon arrival in the United States, and often the intent of the traveler isn’t criminal. With this information in mind,  Rajkumar and his team decided to make a website and mobile app called Safe Souvenirs to educate international travelers about which products will be confiscated if they buy them and later try to travel with them. The website can be accessed from airport kiosks before going through security.

    Rajkumar (far left) and computer science classmate Joshric Aurea (near left) hooked up with UC San Diego alumni Shannon Chamberlin (Ed.D., Teaching and Learning, ’14) and Alicia Johal (BS, Biology, ’10) at the event on Friday night, after learning that Chamberlin’s brother-in-law, also present at the event, had experience with iOS app development.

    “Getting to work alongside industry professionals was really special because we got a different perspective,” said Rajkumar, who previously participated in a student hackathon in 2015. “We wanted to come up with a solution that has impact right off the bat.”

    They’ll have that chance – Rajkumar’s team came in a close second, behind Wild Track, a team consisting of Fab Lab personnel and Accel Robotics co-founder and UC San Diego alumnus Nick Morozovsky (Ph.D., Mechanical Engineering, ’14).
    Morozovsky’s team sought to improve upon a software tool known as Spatial Monitoring and Reporting Tool, or SMART, the most widely adopted solution for managing law enforcement, ecological monitoring and intelligence in protected areas. While SMART does extremely well at capturing the data collected by rangers and conservationists, no solution exists to support local communities in reporting critical information such as poaching incidents, intelligence on poachers, wildlife trafficking and details of human-wildlife conflict.

    "Many residents of protected areas don't have smart phones," said Morozovsky. "We created a system in which a user can report illegal activity by sending an SMS text message to a number we selected, and the content will be analyzed for language and keywords. The data will then be imported into an existing database for law enforcement."

    The judges selected Wild Track as the winning hack because it addressed the problem at a deeper level, is both sustainable and scalable, and user-friendly. Both Wild Track and Safe Souvenirs won a private behind-the-scenes tour of the San Diego Zoo Safari Park.

  • CSE Sophomores Take Top Prize at SD Hacks 2016

    At one of the largest hackathons organized by students, the UC San Diego student-run SD Hacks took place over the Oct. 1-2 weekend, attracting nearly 1,000 student hackers from UC San Diego and universities around the country (after the competition received applications from more than 4,000 students).

    In the highly competitive rivalry, a team of CSE computer-science sophomores took the Best of Show prize in the 36-hour hackathon for their Re:Art app, a filter-style tool inspired by the app Prisma for processing still images, but with Re:Art, the filters apply to video content. "[We] aim to transform a video into an animated artwork by furnishing it with a particular style or pattern of paint chosen by the user," explained the team in their summary posted on DevPost (see link at bottom). In addition to the $1,000 prize, the students received another $1,000 for the Most Commercializable Hack, awarded by the UC San Diego Office of Innovation and Commercialization.

    The winning team consisted of Chen Yang, Zhuojun Chen, Wanze Xie and Jianhan (Joanna) Xu (pictured at right). "This was a great experience," said Yang (at left). "In addition to building a great product, I mastered Python."  Yang and Xu recently signed up to be new tutors for lecturer Rick Ord's CSE 11 course in the fall quarter. "Their web app Re:Art turns videos into art work using deep learning and neural networks," notes Ord (see image below depicting Re:Art photo in the style of impressionist Paul Gauguin). Specifically, the students used the open-source VGG19 neural network model and Torch Framework, while achieving a non-linear video processing result through Open CV (open source computer vision). The team rounded out its trifecta of wins with a prize for Best Use of AWS (Amazon Web Services) on Amazon's EC2 cloud platform.

    The Re:Art teammates were among many CSE students participating in SD Hacks 2016. Others included one of the organizers, junior Yacoub Oulad Daoud, and Andres Gomez, a computer-engineering senior who developed a mobile app to inform incoming freshmen of all the majors available at UC San Diego. Prior to the start of the competition,  CSE alumna Sarah Guthals (B.S. ’10, M.S. ’12, Ph.D. ’14) encouraged students to consider many career paths where computer science is critical. "There are many ways you can use it to contribute to our world," said Guthals. "That doesn't just mean working for a software company."

    Among others, the CSE department was a major sponsor of SD Hacks 2016, as were Qualcomm Institute, the Jacobs School, Rady School of Management, and companies, among them ViaSat, whose executives on hand included CSE alumnus Nikolai Deveraux (B.S. '01), and many others. [Photos by Annie Liou Photography]

    Read the full recap about SD Hacks 2016 on the Jacobs School of Engineering website.
    Learn more about Re:Art and see the app in action on DevPost.

  • Computer Science and Engineering at San Diego Maker Faire 2016

    It was billed as “The Greatest Show (&Tell) on Earth,” and students from UC San Diego were again  part of the spectacle as Maker Faire San Diego took over Balboa Park over the weekend of October 1-2. For the second consecutive year, CSE students were among the participants from the Jacobs School of Engineering, the Qualcomm Institute and their joint Engineers for Exploration student program. One CSE alumnus (at left) , Stephen Foster (Ph.D. '15), who co-created the company ThoughtSTEM to teach coding and other IT skills to youngsters, was among the speakers. Foster has also developed educational games such as CodeSpells and LearnToMod. Foster's presentation? "The Matrix is Here. And You Can Hack It" and how to shape the future of immersive virtual reality.

    The celebration was an all-ages gathering of tech enthusiasts, crafters, educators, tinkerers, hobbyists, engineers, science clubs, authors, artists, students and much more. The Faire is a part of Innovate San Diego, a week-long series of events showcasing innovation in the Cali-Baja region, and it has become part of pop culture – a place for experiential marketing, debuting new technologies and inventions, and celebrating geekdom of every stripe through the global, tech-influenced and do-it-yourself community known as the Maker movement.

    The CSE department showcased the Gadgetron Robot Factory, a hands-on experience that uses a GUI-based programming environment to program configurable robots. “We make it easier for people at all skill levels to build and design robots,” Swanson said. “We are really interested in getting feedback from the community.” Gadgetron users decide what they want the robot to do and how they want it to look. The device does the rest, selecting electronic components to use and generating a blueprint for assembly. (Pictured above: Computer science graduate students Michael Madrid Gonzalez, left, and Jorge Garza show two robots built with the Gadgetron Robot Factory tool.)
    CSE students also participate in Engineers for Exploration, which demonstrated an underwater stereo camera rig, a paraglider, and a radio-collar tracker for tracking wildlife. The Qualcomm Institute's DroneLab as well as the Center of Interdisciplinary Art, Architecture and Archaeology (CISA3) demonstrated several different types of drones taking photos and video to produce compelling imagery resulting from the documentation of historic sites on land and underwater using aerial, terrestrial and ocean-going remote drones.
  • Dog Poop Microbiome Predicts Canine Inflammatory Bowel Disease

    Our gut microbiomes — the varieties of microbes living in our digestive tracts — may play a role in inflammatory bowel disease (IBD). “One of the really frustrating things about IBD in humans is that it’s hard to diagnose — it usually requires intestinal biopsies, which are not only imperfect, but invasive and expensive to collect,” said senior author Rob Knight, PhD, professor in the Departments of Pediatrics as well as Computer Science and Engineering, and director of the Center for Microbiome Innovation, all at UC San Diego.

    Since dogs can also suffer from IBD, researchers at the University of California San Diego School of Medicine analyzed fecal samples from dogs with and without the disease. They discovered a pattern of microbes indicative of IBD in dogs. With more than 90 percent accuracy, the team was able to use that information to predict which dogs had IBD and which did not. However, they also determined that the gut microbiomes of dogs and humans are not similar enough to use dogs as animal models for humans with this disease.

    The study is published October 3 in Nature Microbiology.

    According to Knight, most people with IBD have similar changes in the types of microbes living in their intestinal tracts, relative to healthy people. Yet it’s still difficult to discern healthy people from those with IBD just by looking at the microbes in their fecal samples. In addition, Knight said it’s not yet clear whether the microbial patterns associated with IBD contribute to the disease’s cause or are a result of the disease.

    In a separate line of study, pets appear to be a conduit for microbe sharing in a house. Knight and collaborators previously found that microbial communities on adult skin are on average more similar to those of their own dogs than to other dogs. With a fair amount of precision, they can pick your dog out of a crowd based solely on overlap in your microbiomes.

  • UC San Diego Department and UCTV Launch ‘The Computer Science Channel’

    University of California Television (UCTV) and the Computer Science and Engineering (CSE) department at the University of California San Diego have launched The Computer Science Channel online. The new video channel began service October 1, 2016, with a slate of newly-produced and existing content.

    “Our faculty are directing research that is transforming the world in a variety of ways and have made us one of the top research departments worldwide,” said CSE Chair Dean Tullsen in a welcome video for the channel. “We have a dynamic student population at both the graduate and undergraduate levels, who are making a difference both on campus and in a variety of leadership positions after they graduate.”

    The computer-engineering professor went on to note that computer science is increasingly integrated into our daily lives. Said Tullsen: “This channel is bringing stories about how computer science is revolutionizing many industries and disciplines, from computer graphics to computational biology.”  

    The Computer Science Channel aims to connect viewers to the world of computing and its impact on the world around us.  New content produced for the Channel in a partnership between CSE and UCTV will also begin airing on UCTV and UCSD-TV over cable and online television services such as Roku and Amazon Fire in the near future.

    “UCTV is looking forward to presenting the people and stories from UC San Diego and around the University of California, that make computer science such an important and fascinating field,” says UCTV Director Lynn Burnstan. 

    New video features debuting on the channel in October include a series called Bits & Bytes, a collection of short features (generally from three to 20 minutes in length).  They include a 14-minute report on “Teaching Computer Science Online” – about UC San Diego’s pioneering computer scientists who are in the vanguard of a new movement to offer high-level computer science courses (known as massive open online courses, or MOOCs) to meet some of the pent-up demand for high-level computer science education around the world. So far, over 1,000,000 learners around the world have registered to take the UC San Diego faculty-developed courses via the two largest online education platforms, Coursera and edX. Some of the UC San Diego MOOCs have surpassed 250,000 learners each.

    The report features interviews with CSE professors Pavel Pevzner, Ravi Ramamoorthi, Scott Klemmer, Christine Alvarado and Mia Minnes, as well as Beth Simon, a professor who recently moved from CSE to the Education Studies program to help UC San Diego faculty to adopt some elements of online learning to their courses in the classroom. (Simon spent the last two years on leave at… Coursera.) 

    Bits & Bytes also features a six-minute interview with CSE professor Steven Swanson and students in his course for seniors, dubbed “The Quadcopter Class" (pictured at left). In it, students build tiny quadcopter drones as a capstone project, and in the process get comprehensive experience in conceiving, designing, building and programming a remote-controlled quadcopter.  “Some of us knew how to do the software, some of us knew how to do the hardware. I came in knowing a lot about quadcopters but not too much about the hardware or software behind them,” says undergraduate David Smith. “It’s a really good class to learn something and to do a fun project.”

    The Computer Science Channel will also document where some of the department’s students go after graduation. For the October launch, the Alumni Profiles section poses the question: How does a single mom on her own, far from home, achieve success in the world of computer science? In her own words, CSE alumna Anu Mupparthi (at right)  (B.S. ’08, M.S. ’11) describes the special roles the department and the field of computer science played in her development from single-parent computer novice to software engineer at Google Photos.

    The joint CSE and UCTV channel also features Computing Primetime, a year-old series of long-form programs (30 to 60 minutes in length) about how computer science is interacting with and transforming many other disciplines and sectors.

  • #1 Employer of Recent CNS-Affiliated Graduates? It's Google

    Most of the Ph.D. and M.S. students who worked in the labs of Center for Networked Systems (CNS) member faculty are well-positioned to land a great job after graduation. A few remain in academia, but the vast majority go to jobs in the technology industry, and not just any jobs. According to a survey of 16 CNS-affiliated graduate students who matriculated in 2015-2016, fully half of the mostly Computer Science and Engineering graduates now work for Google, with others going to fast-track jobs at Apple, Facebook, and other tech companies.

    Staying in academia

    Only a few graduating CNS students are staying in academic environments. According to CNS co-director George Porter, “that’s more than 18 percent of our 16 recent graduates, and that’s probably pretty standard among the top 20 schools.”

    After working with his advisor, CSE Prof. YY Zhou, Peng (Ryan) Huang (Ph.D. '16) was offered a tenure-track position as Assistant Professor in computer science at Johns Hopkins University. He was offered the job because the university is trying build a new area of strength in Huang's area – computer systems. “I’m particularly interested in understanding growing problems in real-world systems and reflecting that understanding in new techniques to improve system reliability,” says Huang. In his dissertation, Huang analyzed the distinctive characteristics of failures in industrial-strength cloud systems. (For more on Huang, see full news release here.) 

    [Pictured (l-r): Ryan Huang, Baris Aksanli, and Bharathan Balaji]

    Since graduating in June 2015, Baris Aksanli (Ph.D. ’15) remains affiliated with CNS as a postdoctoral researcher in Tajana Rosing’s lab. Then in August 2016, he became an Assistant Professor in the Electrical and Computer Engineering department of San Diego State University (SDSU), where Aksanli teaches embedded-systems courses and real-time operating systems. On the research side he continues to work on energy efficiency in various domains, including embedded systems, data centers, Internet of Things, and cyber-physical systems. Aksanli did two internships in graduate school, one at Intel (in 2012), the other at Lawrence Berkeley National Laboratory (2011).

    Bharathan Balaji (Ph.D. ’16), who previously received his M.S. in 2011 from the ECE department before transferring to CSE, is also remaining in academia. He recently joined UCLA as a postdoctoral researcher in the lab of Prof. Mani Srivastava, a longtime collaborator of Balaji’s UCSD advisor Rajesh Gupta (and co-advisor Yuvraj Agarwal, now at Carnegie Mellon). Balaji’s research focused on improving energy efficiency of buildings by creating software applications that exploit existing infrastructure to provide services such as information organization, fault detection, personalized control, and sensing when an occupant is in the building. As a grad student, Balaji did an internship at Ericsson Research (working on a Wi-Fi-based occupancy sensing solution).

    Google hires 50% of CNS recent graduates

    Google may be the most sought-after employer of computer science Ph.D. graduates, so it’s perhaps no surprise that Google hired more CNS recent graduates than any other company (all of them armed with degrees from CSE at UC San Diego). Fully half of the graduates – eight of the 16 CNS recent alumni – are now employed at Google.

    On September 15, Wilson Wing-Soon Lian (M.S., Ph.D. ’13, ’16) defends his dissertation on “JIT Spraying Threats on ARM and Defense by Diversification”. Lian’s dissertation committee was co-chaired by his advisors, Stefan Savage and Hovav Shacham. Lian says his research interests are “broadly in security and privacy, but lately I’ve been looking at the security of Just-In-Time compilers.” As a graduate student from 2010 to 2016, he was hired and re-hired at Google for three summer internships in 2012, 2013 and 2015, so it’s no surprise that, with his Ph.D. in sight, Lian has already accepted a job at… Google. He’ll be a full-time software engineer.

    Jagannathan Venkatesh (Ph.D. ‘16) graduated in June after defending his dissertation on “A Context-Aware Approach to Residential Grid Automation.” In his thesis, Venkatesh proposed “using context – additional high-level information – about elements of the smart grid (sources, loads and storage) to improve the efficiency of its operations.” At the all-campus graduation ceremony, his advisor Tajana Rosing was on hand (pictured at right with Venkatesh). Today the CSE and CNS alumnus works at Google, where he had previously done three internships in 2011, 2012 and 2013, including the development of a testing framework for video ads, a tool to search, analyze and debug Google’s social back-end data, and designing user interfaces that are intuitive to users and reusable by developers.

    [Pictured at right (clockwise from top left): Wilson Lian, Jagannathan Venkatesh, Mike Conley, Tristan Halvorson, Rishi Kapoor, Lonnie Liu, Malveeka Tewari and Liqiong Yang.]

    Mike Conley (M.S., Ph.D. ’12, ’15) completed graduate school in computer science under George Porter and Amin Vahdat. His primary research interests were in the areas of big data, I/O-intensive computation, distributed systems, cloud computing, MapReduce, data centers and high-speed sorting. Conley’s doctoral dissertation on "Achieving Efficient I/O with High-Performance Data Center Technologies," focused on the performance of storage and network I/O in large-scale distributed systems (notably on TritonSort and Themis), and he demonstrated how to run such applications on a wide variety of hardware platforms, from solid-state disks to supercomputers. Since October 2015, Conley has been a software engineer at Google in Mountain View, CA, where he also did internships in 2010 and 2011.

    Tristan Halvorson (Ph.D. ’15) studied the domain name market, measuring the market with web and whois data to determine the goal of domain name registrants. Previously with his advisors Stefan Savage and Geoffrey Voelker, Halvorson investigated email spam from a monetary perspective by measuring many email spammers' costs and revenue. He also spent the summer of 2012 on an internship with Yahoo!’s email anti-spam team, with whom he analyzed data on Hadoop to look for compromised webmail accounts.) On graduation, he joined Google as a software engineer.