Entity

Time filter

Source Type

RPI
United States

Freedman D.,RPI | Turek M.W.,Kitware
Image and Vision Computing | Year: 2010

Many problems in computer vision can be posed in terms of energy minimization, where the relevant energy function models the interactions of many pixels. Finding the global or near-global minimum of such functions tends to be difficult, precisely due to these interactions of large (> 3) numbers of pixels. In this paper, we derive a set of sufficient conditions under which energies which are functions of discrete binary variables may be minimized using graph cut techniques. We apply these conditions to the problem of incorporating shape priors in segmentation. Experimental results demonstrate the validity of this approach. © 2009 Elsevier B.V. All rights reserved. Source


Wu Q.,Skyera Inc. | Zhang T.,RPI
IEEE Transactions on Computers | Year: 2013

This paper advocates a time-aware design methodology for using multilevel per cell (MLC) phase-change memory (PCM) in data storage systems such as solid-state disk and disk cache. It is well known that phase-change material resistance drift gradually reduces memory device noise margin and degrades the raw storage reliability. Intuitively, due to the time-dependent nature of resistance drift, if we can dynamically adjust storage system operations adaptive to the time and, hence, memory cell resistance drift, we may improve various PCM-based data storage system performance metrics. Under such an intuitive time-aware system design concept, we propose three specific design techniques, including time-aware variable-strength error correction code (ECC) decoding, time-aware partial rewrite, and time-aware read-&-refresh. Since PCM-based data storage systems have to use powerful ECC whose decoding can be energy-hungry, the first technique aims to minimize the ECC decoding energy consumption. The second technique improves the data retention limit when using partial rewrite in MLC PCM, and the third technique can further improve the efficiency of time-aware variable-strength ECC decoding. Using hypothetical 2-bit/cell PCM with device parameters from recent device research as a test vehicle, we carry out mathematical analysis and trace-based simulations, which show that these techniques can improve the data retention limit by few orders of magnitude, and enable up to 97 and 79 percent energy savings for PCM-based solid-state disk and PCM-based disk cache. © 1968-2012 IEEE. Source


The eigenfunctions of the depth separated wave equation are expanded in terms of a known finite basis set, of size M, with unknown coefficients. The coefficients are found by requiring that the expansion satisfies a variational form of the wave equation, when restricted to the subspace spanned by the basis. This is the Galerkin approximation, wherein one obtains an MxM matrix eigenvalue problem. The convergence of the matrix eigenvalues depends on the suitability of the chosen basis set. Typically, the errors in the matrix eigenvalues are bounded by 1/M∧r, for large M, where the exponent r > 0 is the rate of convergence. Two basis sets are considered: one uses trigonometric functions (Fourier) and the other uses polynomials (Legrendre). The density discontinuity at the bottom of the ocean creates a corner in the eigenfunctions that should be built into the basis sets. A corner in the sound speed profile (e.g., at the bottom of the mixed-layer) yields r = 3, which assures convergence. The distribution of errors is a determining factor in the choice between Fourier-Galerkin and Legendre-Galerkin. The errors in the first 1/3 of the eigenvalues are orders of magnitude smaller with the Legendre-Galerkin method, in the problem presented. © 2013 Acoustical Society of America. Source


News Article
Site: http://phys.org/nanotech-news/

Phosphorene, a single layer of phosphorous in a particular configuration, has potential application in semiconductor transistors. Credit: Matthew Cherny Two-dimensional phosphane, a material known as phosphorene, has potential application as a material for semiconducting transistors in ever faster and more powerful computers. But there's a hitch. Many of the useful properties of this material, like its ability to conduct electrons, are anisotropic, meaning they vary depending on the orientation of the crystal. Now, a team including researchers at Rensselaer Polytechnic Institute (RPI) has developed a new method to quickly and accurately determine that orientation using the interactions between light and electrons within phosphorene and other atoms-thick crystals of black phosphorus. Phosphorene—a single layer of phosphorous atoms—was isolated for the first time in 2014, allowing physicists to begin exploring its properties experimentally and theoretically. Vincent Meunier, head of the Rensselaer Department of Physics, Applied Physics, and Astronomy and a leader of the team that developed the new method, published his first paper on the material—confirming the structure of phosphorene—in that same year. "This is a really interesting material because, depending on which direction you do things, you have completely different properties," said Meunier, a member of the Rensselaer Center for Materials, Devices, and Integrated Systems (cMDIS). "But because it's such a new material, it's essential that we begin to understand and predict its intrinsic properties." Meunier and researchers at Rensselaer contributed to the theoretical modeling and prediction of the properties of phosphorene, drawing on the Rensselaer supercomputer, the Center for Computational Innovations (CCI), to perform calculations. Through the Rensselaer cMDIS, Meunier and his team are able to develop the potential of new materials such as phosphorene to serve in future generations of computers and other devices. Meunier's research exemplifies the work being done at The New Polytechnic, addressing difficult and complex global challenges, the need for interdisciplinary and true collaboration, and the use of the latest tools and technologies, many of which are developed at Rensselaer. In their research, which appears in ACS Nano Letters, the team initially set out to refine an existing technique for determining the orientation of the crystal. This technique, which takes advantage of Raman spectroscopy, uses a laser to measure vibrations of the atoms within the crystal as energy moves through it, caused by electron-phonon interactions. Like other interactions, electron-phonon interactions within atoms-thick crystals of black phosphorus are anisotropic and, once measured, have been used to predict the orientation of the crystal. In reviewing their initial results from Raman spectroscopy, the team noticed several inconsistencies. To investigate further, they obtained actual images of the orientation of their sample crystals using Transmission Electron Microscopy (TEM), and then compared them with the Raman spectroscopy results. As a topographic technique, TEM offers a definitive determination of the orientation of the crystal, but isn't as easy to obtain as the Raman results. The comparison revealed that electron-phonon interactions alone did not accurately predict the orientation of the crystal. And the reason why led the way to yet another anisotropy of phosphorene—that of interactions between photons of light and electrons in the crystal. "In Raman you use a laser to impart energy into the material, and it starts to vibrate in ways that are intrinsic to the material, and which, in phosphorene, are anisotropic," said Meunier. "But it turns out that if you shine the light in different directions, you get different results, because the interaction between the light and the electrons in the material—the electron-photon interaction—is also anisotropic, but in a non-commensurate way." Meunier said the team had reason to believe phosphorene was anisotropic with respect to electron-photon interactions, but didn't anticipate the importance of the property. "Usually electron-photon anisotropy doesn't make such a big difference, but here, because we have such a particular chemistry on the surface and such a strong anisotropy, it's one of those materials where it makes a huge difference," Meunier said. Although the discovery revealed a flaw in the interpretations of Raman spectra relying on electron-phonon interactions, it also revealed that electron-photon interactions alone provide an accurate determination of the orientation of the crystal. "It turns out that it's not so easy to use Raman vibrations to find out the direction of the crystal," Meunier said. "But, and this is the beautiful thing, what we found is that the electron-photon interaction (which can be measured by recording the amount of light absorbed)—the interaction between the electrons and the laser—is a good predictor of the direction. Now you can really predict how the material will behave as a function of excitement with an outside stimulus." Explore further: Electrons move like light in three-dimensional solid More information: Xi Ling et al. Anisotropic Electron-Photon and Electron-Phonon Interactions in Black Phosphorus, Nano Letters (2016). DOI: 10.1021/acs.nanolett.5b04540


RPI researcher Stacy Patterson is developing an easy to use framework for conducting data analytics across multiple IoT devices. Credit: Rensselaer The Internet of Things promises to improve our lives by connecting sensors in the objects that surround us - buildings, appliances, gadgets, and vehicles - and the data that they collect. But to realize that potential, programmers need tools that make it easier to create applications that combine devices and the cloud. With the support of the National Science Foundation (NSF), researcher Stacy Patterson is building those tools and developing a framework that developers can use to easily perform data analytics over a multitude of devices. "Rather than developers designing custom algorithms for each network of devices, we're going to build a framework of software that sits on all these devices and the cloud that will automatically manage communication between the devices and deal with device and network failures," said Patterson, the Clare Boothe Luce Assistant Professor of Computer Science at Rensselaer Polytechnic Institute (RPI). "Now the developer only needs to provide a little bit of code to say 'this is how I want it to work,' and this framework will take care of the rest." The project, "Toward a Machine Learning Framework for the Internet of Things," is supported by a prestigious five-year $618,661 NSF Faculty Early Career Development Award (CAREER). "Dr. Patterson is developing approaches that advance the vision of Internet of Things technology," said Curt Breneman, dean of the Rensselaer School of Science. "This CAREER award recognizes the promise of that work, and we congratulate her for it." Patterson's research epitomizes the work being done at The New Polytechnic, addressing difficult and complex global challenges, the need for interdisciplinarity and true collaboration, and the use of the latest tools and technologies, many of which are developed at Rensselaer. The Internet of Things (IoT) describes a network of devices, from RFID tags, to smart thermostats, to light bulbs, that can sense and communicate information. It is predicted that, by 2020, there will be 25 to 50 billion IoT devices. This massive network and the data it generates will enable new applications in a wide range of critical domains including environmental management, smart infrastructure, and health care. The project is an extension of Patterson's current research into enhancing the utility of sensors embedded in automobiles, by creating real-time networks that allow automobiles to pool their individual information into a larger shared picture of driving conditions in the area. To achieve the vision of IoT, as well as in her work with automobile networks, Patterson said, it is crucial to be able to quickly analyze and learn from the massive amount of generated data. Current approaches for big data analytics require full data transfer to a platform with large computational power, such as the cloud. Given the projected explosion in the number of devices and the resulting data generation rate, this is not feasible. Patterson said the grant will help her achieve three goals in IoT research. The first is to develop a computational framework that reduces the problem to an abstraction, anticipating considerations like the type and quality of data, the number of devices, and how the data are related across devices. "What kinds of relationships are people interested in with this data, and how does that embed down to the physical world?" Patterson said. "This is ultimately about searching for a pattern in how I would solve the kinds of problems that interest people." The second goal is to provide a stable platform that masks the differences between devices, and compensates for failed devices or computers and lost data. Finally, she will build tools to enforce a standard for speed and accuracy of the framework. Explore further: Managing the “Internet of Things

Discover hidden collaborations