UCSD Department of Computer Science and Engineering

Friday, November 21, 2008
TALK: Vish Krishnan

PDF Requires Acrobat Reader

CSE Server/Network Status

A Picture Is Worth A Thousand Locksmiths

UC San Diego computer scientists have built a software program that can perform key duplication without having the key. Instead, the computer scientists only need a photograph of the key.

The bumps and valleys on your house or office keys represent a numeric code that completely describes how to open your particular lock. If a key doesn't encode this precise "bitting code," then it won't open your door.

"We built our key duplication software system to show people that their keys are not inherently secret," said Stefan Savage, the computer science professor from UC San Diego's Jacobs School of Engineering who led the student-run project. "Perhaps this was once a reasonable assumption, but advances in digital imaging and optics have made it easy to duplicate someone's keys from a distance without them even noticing."

Professor Savage presented this work on October 30 at ACM's Conference on Communications and Computer Security (CCS) 2008, one of the premier academic computer security conferences.

The keys used in the most common residential locks in the United States have a series of 5 or 6 cuts, spaced out at regular intervals. The computer scientists created a program in MatLab that can process photos of keys from nearly any angle and measure the depth of each cut. String together the depth of each cut and you have a key's bitting code, which together with basic information on the brand and type of key you have, is what you need to make a duplicate key.

The chief challenge for the software system, called "Sneakey", is to adjust for a wide range of different angles and distances between the camera and the key being captured. To do so, the researchers relied on a classic computer vision technique for normalizing an object's orientation and size in three dimensions by matching control points from a reference image to equivalent points in the target image.

Good Code, Bad Computations
Two graduate students from UC San Diego's computer science department--Erik Buchanan and Ryan Roemer--have just published work showing that the process of building bad programs from good code using "return-oriented programming" can be automated and that this vulnerability applies to RISC computer architectures and not just the x86 architecture (which includes the vast majority of personal computers). Last year, UC San Diego computer science professor Hovav Shacham formally described how return-oriented programming could be used to force computers with the x86 architecture to behave maliciously without introducing any bad code into the system. However, the attack required painstaking construction by hand and appeared to rely a unique quirk of the x86 design.

UCSD Ranked Among the Nation's Best Values for a College Education
UC San Diego is ranked by Kiplinger's Personal Finance magazine in the top 10 among the best values for public colleges and universities in the United States. The school was ranked 10th in value for in-state costs and seventh for expenses for students from out of the state. The editors said they took into account academic quality, admission rates, student-faculty ratios, graduation rates, costs and availability of financial aid.

When You Look at a Face, You Look Nose First
While general wisdom says that you look at the eyes first in order to recognize a face, UC San Diego computer scientists now report that you look at the nose first. The nose may be the where the information about the face is balanced in all directions, or the optimal viewing position for face recognition, the researchers from Temporal Dynamics of Learning Center propose in a paper recently published in the journal Psychological Science. The researchers showed that people first look just to the left of the center of the nose and then to the center of the nose when trying to determine if a face is one they have seen recently. These two visual fixations near the center of the nose are all you need in order to determine if a face is one that you have seen just a few minutes before. Looking at a third spot on the face does not improve face recognition, the cognitive scientists found.

UC Researchers Explore Tactile Feedback for Cell Phones
Researchers from UC San Diego and Microsoft presented a paper last week that explores how the next generation of cell phones or computer games might employ new forms of "vibrotactile feedback" to improve communications between people and machines. "People don't go around vibrating each other. We have a much larger tactile vocabulary for human-to-human communications, which we think can be expanded to devices," said Kevin Li, 26, a doctoral candidate in UCSD's department of computer science and engineering and lead author of the paper "Tapping and Rubbing: Exploring New Dimensions of Tactile Feedback with Voice Coil Motors."

UC San Diego Proteomics Pioneers Establish First Center Dedicated to Computational Mass Spectrometry
UC San Diego engineers and scientists have received a five-year $4.94M grant from the National Center for Research Resources (NCRR), a part of the National Institutes of Health (NIH), to develop algorithms and software for deciphering all the proteins that are present in biological samples. This "proteomics" work promises to revolutionize routine blood tests, vaccine development, cancer diagnostics, and many other important biomedical challenges, says Pavel Pevzner, the UC San Diego Jacobs School of Engineering computer science professor leading the project. The new grant will also support development of the software infrastructure required to share these cutting edge computational mass spectrometry tools with researchers around the nation and the world. This effort will combat a global computational bottleneck that is currently holding back the field of proteomics, which by definition strives to glean biological insights from looking at all the proteins present in biological samples. While there are traditional tools to do some of this proteomics work, they are time consuming and expensive, and have contributed to the computational bottleneck.

9500 Gilman Drive, La Jolla, CA 92093-0404