| Name: | Basharat Amna |
| E-mail: | amna.basharat@gmail.com |
| Institute: | UGA |
| Poster Title: | |
| Poster Abstract: | |
| Name: | David Bader |
| E-mail: | bader@cc.gatech.edu |
| Institute: | Georgia Institute of Technology |
| Plenary Talk Title: | OPPORTUNITIES AND CHALLENGES IN MASSIVE DATA-INTENSIVE COMPUTING |
| Plenary Talk Abstract: | Emerging real-world graph problems include detecting community structure in large social networks, improving the resilience of the electric power grid, and detecting and preventing disease in human populations. Unlike traditional applications in computational science and engineering, solving these problems at scale often raises new challenges because of sparsity and the lack of locality in the data, the need for additional research on scalable algorithms and development of frameworks for solving these problems on high performance computers, and the need for improved models that also capture the noise and bias inherent in the torrential data streams. In this talk, the speaker will discuss the opportunities and challenges in massive data-intensive computing for applications in computational biology, genomics, and security. The explosion of real-world graph data poses a substantial challenge: How can we analyze constantly changing graphs with billions of vertices? Our approach leverages fine-grained parallelism, lightweight synchronization, and shared memory, to scale to massive graphs. |
| Name: | Michele Benzi |
| E-mail: | benzi@mathcs.emory.edu |
| Institute: | Emory University |
| Plenary Title: | Ranking Hubs and Authorities Using Matrix Functions |
| Plenary Abstract: | The notions of subgraph centrality and communicability, based on the exponential of the adjacency matrix of the underlying graph, have been effectively used in the analysis of undirected networks. In this talk we propose an extension of these measures to directed networks, and we apply them to the problem of ranking hubs and authorities. The extension is achieved by bipartization, i.e., the directed network is mapped onto a bipartite undirected network with twice as many nodes in order to obtain a network with a symmetric adjacency matrix. We explicitly determine the exponential of this adjacency matrix in terms of the adjacency matrix of the original, directed network, and we give an interpretation of centrality and communicability in this new context, leading to a technique for ranking hubs and authorities. The matrix exponential method for computing hubs and authorities is compared to the well known HITS algorithm, both on small artificial examples and on more realistic real-world networks. This is joint work with Christine Klymko (Emory) and Ernesto Estrada (Strathclyde). |
| Name: | Sebastian Berisha |
| E-mail: | sberish@emory.edu |
| Institute: | Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Luca Bertagna |
| E-mail: | luca.bertagna84@gmail.com |
| Institute: | Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Qing Chu |
| E-mail: | qchu@emory.edu |
| Institute: | Emory University |
| Poster Title: | ITERATIVE WAVEFRONT RECONSTRUCTION IN ADAPTIVE OPTICS |
| Poster Abstract: | Obtaining high resolution images of space objects from ground based telescopes is challenging, and often requires computational post processing using image deconvolution methods. Good reconstructions can be obtained if the convolution kernel can be accurately estimated. The convolution kernel is defined by the wavefront of light, and how it is distorted as it propagates through the atmosphere. In this paper we describe the wavefront reconstruction problem, and more specifically, a new linear least squares model that exploits information from multiple measurements. |
| Name: | Luca Dieci |
| E-mail: | dieci@math.gatech.edu |
| Institute: | |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Xiaochen Dong |
| E-mail: | xxdong@uga.edu |
| Institute: | UGA |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Sung Ha Kang |
| E-mail: | kang@math.gatech.edu |
| Institute: | Georgia Tech |
| Plenary Title: | Variational models for image segmentation and multiphase extensions |
| Plenary Abstract: | This talk is on variational approaches to image reconstruction and segmentation. We consider image segmentation problems starting from Mumford-Shah and Chan-Vese model. The main focus will be on unsupervised model and multiphase segmentation. Some extensions including data clustering extension will be presented. |
| Name: | Caner Kazanci |
| E-mail: | caner@uga.edu |
| Institute: | |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Christine Klymko |
| E-mail: | cklymko@emory.edu |
| Institute: | Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Ming-Jun Lai |
| E-mail: | mjlai@math.uga.edu |
| Institute: | University of Georgia |
| Poster Title: | An Oxygen Depletion Estimate for BP Oil Disaster in the Gulf of Mexico |
| Poster Abstract: | I will use bivariate splines to build a model for estimating the oxygen depletion based on the data collected by the research vessel after the disaster of the BP Deep Horizon Exploration. |
| Name: | Hui Liu |
| E-mail: | hliu16@student.gsu.edu |
| Institute: | Georgia State University |
| Poster Title: | A Posteriori Error Analysis for Nonlinear Magnetometry Equation (joint with Alexandra Smirnova) |
| Poster Abstract: | We consider the possibility of {\it a posteriori} error estimates for nonlinear ill-posed operator equations. Given an auxiliary finite-dimensional problem $\Phi(w)=0$, $\Phi:{\mathcal{D}}_\Phi \subset {\mathcal{E}}_N\to {\mathcal{E}}_M$, that approximates the original infinite model $F(x)=0,$ $ F:{\mathcal{D}}_F \subset \mathcal{X}\to \mathcal{Y}$ with a certain level of accuracy, we try to estimate the distance between $z$, an approximate solution to $\Phi(w)=0$, and $x^*$, the exact solution to $F(x)=0$. The problem $\Phi(w)=0$ is assumed to accumulate different sources of error (discretization, measurements, etc), and the computed solution $z$ is assumed to satisfy the equation $\Phi(w)=0$ within a nonzero tolerance $\delta.$ Both theoretical and numerical study of {\it a posteriori} error analysis is conducted. |
| Name: | Tianming Liu |
| E-mail: | tliu@cs.uga.edu |
| Institute: | University of Georgia |
| Plenary Title: | Neuromediomics: the interface of brain imaging and multimedia |
| Plenary Abstract: | The multimedia analysis community has long attempted to bridge the significant gaps between low-level features and high-level semantics for the purpose of content-based multimedia management. However, despite remarkable progress in this field, the gaps are still significant, and advancement of this field entails novel semantics representations that can be effectively linked to low-level multimedia streams. Essentially, from the brain science perspective, the semantics are represented and understood by the human brain, and modern neuroscience has made significant progress in understanding semantics representations in the brain. In particular, functional magnetic resonance imaging (fMRI) has been proven to be a superior, powerful technique to noninvasively probe the brain’s perception and cognition. This talk will introduce our recent effort in marrying these two fields in order to bridge the significant gaps between low-level multimedia features and high-level semantics perceived by the human brain. Specifically, we performed fMRI imaging to monitor the brain’s responses under the natural stimulus of movie watching, from which we learned the most descriptive low-level features that best correlate with the fMRI-derived semantic features and learned predictive models that can estimate the brain’s responses by the selected low-level features. In the long-run, we hope that our work will facilitate wider adoption of applying fMRI to assist multimedia analysis and using multimedia as fMRI natural stimuli to better understand the brain. |
| Name: | Jun Lu |
| E-mail: | jlu39@math.gatech.edu |
| Institute: | Georgia Institute of Technology |
| Poster Title: | A fast algorithm for finding the shortest path by solving initial value ODEs |
| Poster Abstract: | We propose a new fast algorithm for finding a global shortest path connecting two points while avoiding obstacles in a region by solving an initial value problem of ordinary differential equations under random perturbations. The idea is based on the fact that every shortest path possesses a simple geometric structure. This enables us to restrict the search in a set of feasible paths that share the same structure. The resulting search set is a union of sets of finite dimensional compact manifolds. Then, we use a gradient flow, based on an intermittent diffusion method in conjunction with the level set framework, to obtain global shortest paths by solving a system of randomly perturbed ordinary differential equations with initial conditions. Comparing to the existing methods, such as the combinatorial methods or partial differential equation methods, our algorithm seems to be faster and easier to implement. We can also handle cases in which obstacle shapes are arbitrary and/or the dimension of the base space is three or higher. |
| Name: | Mirabella Lucia |
| E-mail: | lucia.mirabella@bme.gatech.edu |
| Institute: | Georgia Institute of Technology |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Leopold Matamba Messi |
| E-mail: | lmatamba@math.uga.edu |
| Institute: | The University of Georgia |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Leslie Meadows |
| E-mail: | lmeadows2@gsu.edu |
| Institute: | Georgia State University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Veronica Mejia Bustamante |
| E-mail: | vmejia@emory.edu |
| Institute: | Emory University |
| Poster Title: | Implementation of Iterative Solvers for the Digital Tomosynthesis Problem in GPUs |
| Poster Abstract: | Tomosynthesis imaging provides a viable alternative to computed tomography (CT) and has obtained significant interest from the medical community as a means for diagnostic radiology and radiation therapy. In digital tomosynthesis imaging, multiple projections of an object are obtained along a small range of different incident angles in order to reconstruct a 3D representation of the object. In this paper we discuss the implementation details of the polyenergetic digital breast tomosynthesis reconstruction algorithm in a GPU using OpenCL. We describe three different algorithm implementations: a serial implementation, a GPU implementation threaded by functionality of the model, and a GPU fused kernel implementation which is threaded to increase performance, throughput, and GPU utilization in the application. We show that the explicit kernel fusion achieves significant speed-up in the reconstruction process of a clinical size patient data set, from running over 100X faster than the version threaded by functionality to 200X faster than the serial approach. |
| Name: | James Nagy |
| E-mail: | nagy@mathcs.emory.edu |
| Institute: | Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Tiziano Passerini |
| E-mail: | tiziano@mathcs.emory.edu |
| Institute: | Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | George Slavov |
| E-mail: | gpslavov@uga.edu |
| Institute: | University of Georgia |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Alexandra Smirnova |
| E-mail: | asmirnova@gsu.edu |
| Institute: | Georgia State University |
| Poster Title: | A Posteriori Error Analysis for Nonlinear Magnetometry Equation (joint with Hui Liu) |
| Poster Abstract: | We consider the possibility of {\it a posteriori} error estimates for nonlinear ill-posed operator equations. Given an auxiliary finite-dimensional problem $\Phi(w)=0$, $\Phi:{\mathcal{D}}_\Phi \subset {\mathcal{E}}_N\to {\mathcal{E}}_M$, that approximates the original infinite model $F(x)=0,$ $ F:{\mathcal{D}}_F \subset \mathcal{X}\to \mathcal{Y}$ with a certain level of accuracy, we try to estimate the distance between $z$, an approximate solution to $\Phi(w)=0$, and $x^*$, the exact solution to $F(x)=0$. The problem $\Phi(w)=0$ is assumed to accumulate different sources of error (discretization, measurements, etc), and the computed solution $z$ is assumed to satisfy the equation $\Phi(w)=0$ within a nonzero tolerance $\delta.$ Both theoretical and numerical study of {\it a posteriori} error analysis is conducted. |
| Name: | Andrew Sornborger |
| E-mail: | ats@math.uga.edu |
| Institute: | University of Georgia |
| Plenary Talk Title: | Mean, Covariance and Variance in Neural Processes: a New Causal Decomposition for Multivariate Data |
| Plenary Talk Abstract: | In modern neural imaging experiments, gigabytes of multivariate data are acquired in minutes. Typically, dimensional reduction methods must be used in order to make the data tractable. Although neuronal models have causal structure, standard data reduction methods such as the singular value decomposition (SVD), independent component analysis (ICA) or non-negative matrix factorization (NMF) only make use of information at zero-lag. In this talk, I will present a new non-parametric decomposition that makes use of causal information to improve estimates of spatial structure in multivariate imaging data. The new decomposition applies multitaper spectral methods to the statistical detection and estimation of significant causal structures in both the time-dependent mean signal and the covariance of the background stochastic signal latent in neural imaging data. We will compare results from standard methods and from our new method demonstrating how it has advanced our understanding of seizure-related calcium activity in the larval zebrafish. |
| Name: | Thiab Taha |
| E-mail: | thiab@cs.uga.edu |
| Institute: | UGA |
| Poster Title: | NUMERICAL METHODS for 1+2 DIMENSIONAL NONLINEAR SCHRÖDINGER TYPE EQUATIONS |
| Poster Abstract: | The nonlinear Schrödinger equation is of tremendous interest in both theory and applications. Various regimes of pulse propagation in optical fibers are modeled by some form of the nonlinear Schrödinger equation. In this paper we introduce sequential and parallel numerical methods for numerical simulations of the 1+ 2 dimensional nonlinear Schrödinger type equations. The parallel methods are implemented on the multiprocessor system at the University of Georgia(UGA). Our preliminary numerical results have shown that these methods give good results and considerable speedup. |
| Name: | Uthayasanker Thayasivam |
| E-mail: | uthaya@uga.edu |
| Institute: | UGA |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Alessandro Veneziani |
| E-mail: | ale@mathcs.emory.edu |
| Institute: | Department of Mathematics and Computer Science, Emory University |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Xiaolin Wang |
| E-mail: | xwang310@math.gatech.edu |
| Institute: | Georgia Institute of Technology |
| Poster Title: | a numerical study on vorticity-enhanced heat transfer |
| Poster Abstract: | |
| Name: | Xiaofeng Xu |
| E-mail: | xuxiaofeng1989@gmail.com |
| Institute: | |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Xiaojing Ye |
| E-mail: | xiaojing.ye@math.gatech.edu |
| Institute: | |
| Poster Title: | |
| Poster Abstract: | |
| Name: | Haomin Zhou |
| E-mail: | hmzhou@math.gatech.edu |
| Institute: | School of Mathematics, Georgia Tech |
| Poster Title: | |
| Poster Abstract: | |