“When in 2012 a computer learned to recognize cats in YouTube videos and just last month another correctly captioned a photo of “a group of young people playing a game of Frisbee,” artificial intelligence researchers hailed yet more triumphs in “deep learning,” the wildly successful set of algorithms loosely modeled on the way brains grow sensitive to features of the real world simply through exposure.
Using the latest deep-learning protocols, computer models consisting of networks of artificial neurons are becoming increasingly adept at image, speech and pattern recognition — core technologies in robotic personal assistants, complex data analysis and self-driving cars. But for all their progress training computers to pick out salient features from other, irrelevant bits of data, researchers have never fully understood why the algorithms or biological learning work.”
The new work, completed by Hariri Institute Fellow, Pankaj Mehta of Boston University, and David Schwab of Northwestern University, demonstrates that a statistical technique called “renormalization,” which allows physicists to accurately describe systems without knowing the exact state of all their component parts, also enables the artificial neural networks to categorize data as, say, “a cat” regardless of its color, size or posture in a given video.
Pankaj Mehta presented these results at BU’s Hariri Institute as his Junior Fellow Award Lecture.
December 1, 2014 from StateTech
A project team of academia and industry experts is making headway on a multimillion-dollar cloud computing initiative announced by Massachusetts Gov. Deval Patrick in April.
If all goes as planned, the three-year project, known as the Massachusetts Open Cloud (MOC), will pave the way for cloud consumers to customize infrastructure and platform services to best meet their needs. Patrick is hopeful that MOC’s public cloud computing infrastructure will spur Big Data innovation in the state.
Azer Bestavros, director of the Rafik B. Hariri Institute for Computing and Computational Science & Engineering at Boston University, likens the new model to a shopping mall, where customers can choose from multiple retailers and mix and match products and services. In the cloud computing world, this model doesn’t exist, Bestavros says. “All you have are stores like Walmart. They own and operate everything.”
Boston University leads the MOC project team. The MOC project received $3 million in state funding, according to Bestavros. “We have commitments of up to $20 million from universities and industry.”
MOC is an architecture and a model for other governments, he adds. It functions similar to a shopping mall in that technologies from multiple MOC service providers will be located in the same location or data center. Ideally, MOC won’t be restricted to government customers. The general public and companies will be able to buy cloud services using an online user interface, much like the one offered by Amazon Web Services.
Cisco, Intel, Red Hat and Juniper Networks are among the growing number of companies partnering with the MOC project. Academia partners include Northeastern University, University of Massachusetts, Harvard University and the Massachusetts Institute of Technology.
Jan Mark Holzer, a senior consulting engineer at Red Hat, says his company is helping MOC build the cloud infrastructure around OpenStack. Red Hat has moved some hardware into the MOC-designated data center and will continue to install hardware and software for at least the next six months. Hozer stressed that the MOC and its member partners are not interested in promoting a particular vendor or product. It’s an open project.
Hozer says he is hopeful that MOC will be open for research use fairly soon, which could mean months, rather than years.
It’s too early to say how much the MOC services will cost, but the price points have to be competitive because agencies aren’t required to use the services.
Tech professionals who work at election offices have pretty full calendars these days; but Albert Grimes, CIO at the Massachusetts Office of Campaign & Political Finance, took time out of his busy schedule to chat about how the office handles the flood of political contribution data that comes in shortly before Election Day. Grimes spoke at Boston University’s Hariri Institute for Computing and Computational Science & Engineering along with Paul Clark, a data analyst and disclosure business architect with the Federal Elections Commission.
Commenting on this event, ComputerWorld Sharon Machlis notes:
One key, he says, is making sure that people who want to know who’s financing the various candidates have easy access to that data. That constituency includes voters, the media and political enthusiasts, as well as candidates and their organizations.
What’s the most interesting thing he and his team are working on? Data visualization, Grimes said, so it’s easier for people to understand trends in where money is coming from to back various politicians and policies. It’s one thing to get a whole series of reports back on who’s contributed to a politician or political committee, he noted.
November 10 2014 from BU today
BU plans to hire up to six data scientists—the intellectual miners extracting applicable information from the mountains of Big Data—over the next three years
Data scientists use mathematical models to analyze voluminous data and draw knowledge from it that can be used in a variety of applications, from health care and business to design and communications. The University provost’s faculty hiring initiative aims to bolster BU’s ranks in a burgeoning field and also to advance the University’s focus on interdisciplinary research.
“It’s not like we don’t do data science. But we don’t have enough data science scholars to address the need for data science by lots of other disciplines,” says Azer Bestavros, director of BU’s Rafik B. Hariri Institute for Computing and Computational Science & Engineering. Current data science is done by handfuls of professors in “small islands of collaborations,” he says, “two or three faculty working together.…We need to do a lot more.”
Azer Bestavros, Institute Founding Director and professor of computer science, was quoted in a Boston Globe article discussing TripAdvisor’s decision to personalize travel recommendations for its users:
Amazon Inc. pioneered personalized searches more than a decade ago, and now “it’s hard to find a website that doesn’t do it,” said Azer Bestavros, director of the Rafik B. Hariri Institute for Computing and Computational Science & Engineering at Boston University.
Bestavros said TripAdvisor probably spent a lot of time trying to understand the context of each click users made, and how to use that information to improve browsing and search results.
“It’s a big undertaking because of the market research they would have to do,” Bestavros said. “What is the impact on how people use the site?”
The inaugural BU Initiative on Cities (IoC) Urban Seminar, The Open City, examined a key question facing urban communities: How can cities take the massive influx of raw data, turn it into knowledge, and turn that knowledge into a better city?
Speakers from Boston University, Northeastern University and the City of Boston, explored the ways in which big data can be leveraged to improve the very communities from which it originates. Sucharita Gopal (Professor of Earth & Environment, BU), Nigel Jacob (Executive in Residence at Boston University; Co-Founder, New Urban Mechanics), and Dan O’Brien (Director of Research at the Boston Area Research Initiative and Assistant Professor at Northeastern), participated in a moderated discussion led by Seminar Co-Chair Nathan Philips.
Boston University’s Rafik B. Hariri Institute for Computing and Computational Science & Engineering today announced it has received funding from the National Science Foundation (NSF) to develop a “smart-city” cloud platform designed to streamline and strengthen multiple municipal functions. Called SCOPE (A Smart-city Cloud-based Open Platform & Ecosystem), the project will pursue research that aims to improve support for and catalyze innovative solutions for transportation, energy, public safety, asset management, and social services in the City of Boston and across Massachusetts.
Thirty-nine BU students, from math, computer science and entrepreneurship “trekked” to local big data companies to meet with company founders, hear from leading data scientists, connect with other students, and experience incubation lab environment. Organized by the Massachusetts Technology Collaborative, BU students joined close to 100 other juniors, seniors, and grad students from leading computer science, math and business programs in Massachusetts,. Read the the Boston Business Journal article on the event. With a long wait list, MassTech is considering a second week of trekking in October.
Event date: September 18, 2014 5:30-6:45 PM ET
Tsai Performance Center, 685 Commonwealth Ave
The first annual Gerald and Deanne Gitner Family College of Arts & Sciences Lecture will feature Hariri Junior Faculty Fellow and CS Assistant Professor Sharon Goldberg. The lecture is about ”Advancing the Human Condition: An Agenda for Research and Education” and will feature a panel of CAS experts.
This event is free and open to the public, no registration required. If you are interested in other events happening during Alumni Weekend 2014, click here for details.
Rosella Capella, Department of Political Science
Sharon Goldberg, Department of Computer Science
Joseph Harris, Department of Sociology
Lucy Hutyra, Department of Earth & Environment
Jeremy Menchik, Pardee School of Global Studies
Henrik Selin, Pardee School of Global Studies
Virginia Sapiro, Dean of the College of Arts & Sciences
Adil Najam, Dean of the Pardee School of Global Studies
Anthony Janetos, Director of the Pardee Center for the Study of the Longer-Range Future
More information, click here or check the calendar please.
Boston University and University of Pittsburg researchers are utilizing Big Data to find urban walking paths that factor in both distance and safety. Esther Galbrun, a postdoc at BU working with Hariri Faculty Fellow Evimaria Terzi of the Computer Science Department, presented a paper titled “Safe Navigation in Urban Environments” at the 3rd International Workshop on Urban Computing in conjunction with KDD2014. She described a set of algorithms that analyzes the shortest path, the safest path, and alternatives in between to provide options for a person walking through a city while addressing concerns that people face everyday. As cities release more data and researchers like Galbrun find new ways to analyze it, people living in cities can potentially enjoy an improved quality of life.