Computer Science Department News
Winners of Multidisciplinary Research Seed Grants Announced, Computer Science Faculty Members Shuiwang Ji, Andrey Chernikov, Nikos Chrisochoides and Yaohang Li named.
"A Multidisciplinary Approach for Gene Expression Pattern Image Analysis," Shuiwang Ji, Department of Computer Science (PI). Consultants are Andrey Chernikov and Nikos Chrisochoides, Department of Computer Science; Christopher Osgood, Department of Biological Sciences; Sudhir Kumar, School of Life Sciences, Arizona State University; and Patric Lundberg, Microbiology and Molecular Cell Biology, Eastern Virginia Medical School.
The researchers propose to initiate the development of a set of computational methods to automate the analysis of gene expression pattern images in the fruit fly, a canonical model organism. Currently, although a large number of gene expression pattern images has been made available in Drosophila, it is still a standard practice to analyze these images by visual inspection. This manual practice fails to provide an unbiased and systematic comparative analysis, hindering the pace of biological discovery. The objective of this multidisciplinary seed grant is to initiate a long-term project aimed ultimately at building a complete computational pipeline for the analysis of gene expression pattern images in the fruit fly.
This set of methods includes techniques and algorithms for image processing, segmentation, registration, machine learning, data mining and knowledge discovery, and a test bed for result evaluation and interpretation.
Excerpt from the main article follows:
"Toward Solutions to Big Data Challenges in Multiple Disciplinary Applications," Yaohang Li, Department of Computer Science (PI). Other investigators are Duc T. Nguyen, Department of Civil and Environmental Engineering; Masha Sosonkina, Department of Modeling, Simulation and Visualization Engineering; and Jin Wang (consultant), Department of Mathematics and Statistics.
Recent years have witnessed a dramatic increase of data in many fields of science and engineering, due to the advancement of sensors, mobile devices, biotechnology, digital communication and Internet applications. These massive, continuing growing, complex, diverse, distributed data sets are referred to as the "big data." Big data touches every aspects of our life. On one hand, big data provides a rich information source to enable us to gain important insight into various scientific and engineering domains at a scale and level that have never been possible before. Successfully addressing the big data challenge can lead to broad scientific and economic impacts. On the other hand, the growth of big data has outpaced our capability to process, analyze and understand these data sets. Most traditional data processing approaches have failed to scale to big data.
In this proposal, an interdisciplinary team of ODU faculty members plans to tackle the big data challenge by targeting the kernel "big matrix" problem. Instead of the traditionally considered big matrices, typically ranging from hundred by hundred to thousand by thousand, the team will target the big data matrices at the scale of million by million or even billion by billion. One aim will be to develop novel computational methods for low-rank approximation of big matrices, solving extremely large systems of linear equations, and big matrix processing by taking advantage of the most advanced high-performance computing architectures. Another will be to apply the big data solutions that are found to several real-life big data applications, including protein structure prediction, coastal circulation and storm surge finite element models, nuclear physics and tumor modeling.
Full Article text can be found here.