Resume

• 2005

  University of Southern California

  Worked on a DARPA funded program called ACIP.

  University of Southern California

• 2002

  Doctor of Philosophy (PhD)

  Computer Engineering

  IEEE

  ACM

  MSA

  ECEGSA

  University of Maryland

• 2000

  ACM

  IEEE

  ASEE

  Vice Chair Arkansas River Valley IEEE Section

  English

  Arabic

  Master of Science (M.Sc.)

  Computer Engineering

  IEEE

  ACM

  MSA

  ECEGSA

  University of Maryland College Park

• 1999

  Howard University

• 1996

  Information Technology Institute (ITI)

  George Washington University

  Arkansas Tech University

  Los Alamos

  NM

  Joint Faculty

  Los Alamos National Laboratory

  Las Cruces

  New Mexico Area

  Teaching Computer Engineering at the undergraduate and graduate levels.\nConduct research in computer architecture and high performance computing.

  Assistant Professor of Electrical and Computer Engineering

  New Mexico State University

  Los Alamos

  NM

  Collaborated with the Ultrascale Systems Research Center researchers on the Performance Prediction Toolkit (PPT) and contributed to the validation and improvement of the hardware models.

  Research Scientist

  New Mexico Consortium

  Information Technology Institute (ITI)

  Los Alamos National Laboratory

  Los Alamos

  NM

  Mentored and worked with scientists and student interns at Los Alamos on Performance Modeling and Prediction.

  Faculty Mentor

  Russellville

  AR

  Teach Computer Engineering courses

  advise students

  service for the department and University

  develop proposals for funding.

  Assistant Professor of Electrical Engineering

  Arkansas Tech University

  HPCL

  GWU Virginia Campus

  Ashburn

  VA

  Conduct research

  publish manuscripts

  advise students

  contribute to white papers in response to funding opportunities

  Lead Research Scientist

  George Washington University

  Los Alamos

  NM

  Collaborated with the Ultrascale Systems Research Center researchers on the Performance Prediction Toolkit (PPT) and contributed to the validation and improvement of the hardware models for GPUs and CPUs.

  Visiting Research Scientist

  New Mexico Consortium

  Teaching Computer Engineering/Science courses\nConducting Research on Computer Architecture

  University of Maryland

  Diploma

  Computer Software Engineering

  Information Technology Institute

  Giza

  Egypt

• 1991

  Bachelor of Science (B.Sc.)

  Electrical

  Electronics and Communications Engineering

  IEEE

  ACM

  Mansoura University

• Klipsch School of Electrical and Computer Engineering

  Welcome to the Klipsch School of Electrical and Computer Engineering. We are housed in the College of Engineering at NMSU

  which is ranked 10th in the nation for total research and development expenditures in engineering-related projects by the National Science Foundation. In addition

  the College of Engineering is ranked 62nd in the nation by U.S.

  High Performance Computing Lab @ GWU

  The George Washington University High Performance Computing Lab

  LaTeX

  Computer Architecture

  Computer Science

  Microprocessors

  C

  Parallel Computing

  Software Engineering

  Computer Hardware

  C++

  University Teaching

  College Teaching

  Matlab

  Programming

  Scholarship of Teaching and Learning

  Algorithms

  Research

  Medical Imaging

  Machine Learning

  Fortran

  Computer Engineering

  Pre-CAD Normal Mammogram Detection using Gray Level Co-occurrence Matrix Features

  The idea is based on two concepts: first

  designing a \"pre-CAD\"system for detecting only normal mammograms. Those normal mammograms are supposed to be screened-out

  leaving the remaining mammograms that were not detected as normal to the radiologist and legacy/conventional CAD systems for further investigation. The second concept provides duality to the \"pre-CAD\"system by separating the mammograms into two different categories according to their tissue type (i.e.

  fatty or dense)

  and studying each category individually. This separation will enhance the performance of the overall \"pre-CAD\"system

  since the classification of normal and not normal mammograms will be within the same tissue-type group.

  Pre-CAD Normal Mammogram Detection using Gray Level Co-occurrence Matrix Features

  Shuai also presented this paper at the SPIE Photonics West Conference in San Francisco. \n\nIn this paper we benchmark various interconnect technologies including electrical

  photonic

  and plasmonic options. We contrast them with hybridizations where we consider plasmonics for active manipulation devices

  and photonics for passive propagation integrated circuit elements

  and further propose another novel hybrid link that utilizes an on chip laser for intrinsic modulation thus bypassing electro-optic modulation. Link benchmarking proves that hybridization can overcome the hortcomings of both pure photonic and plasmonic links. We show superiority in a variety of performance parameters such as point-to-point latency

  energy efficiency

  capacity

  ability to support wavelength division multiplexing

  crosstalk coupling length

  bit flow density and Capability-to-Latency-Energy-Area Ratio.

  Low latency

  area

  and energy efficient Hybrid Photonic Plasmonic onchip Interconnects (HyPPI)

  X-ray mammograms are one of the most common techniques used by radiologists for breast cancer detection and diagnosis. Early detection is important

  which raised the importance of developing Computer-Aided Detection and Diag-nosis(CAD) systems. Although most(CAD)systems were designed to help radiologists in their diagnosis by providing useful insight

  the accuracy of CAD systems remains below the level that would lead to an improvement in the overall radiologists' performance. Unlike other CAD systems who aim to detect abnormal mammograms

  we are designing a pre-CAD system that aims to detect normal mammograms instead of abnormal ones. The pre-CAD system works as a \"first look\" and screens-out normal mammograms

  leaving the radiologists and other conventional CAD systems to focus on the suspicious cases. Support Vector Machine classifiers are used to detect normal mammograms. We are comparing the effect of using 1-class and 2-class SVMs when normal mammogram

  instead of abnormal

  is detected. Results showed that our pre-CAD system performance for 1-class outperformed 2-class SVM classifiers almost always. Using our set of features

  1-class SVM achieved a specificity of (99.2%)

  while the two-class SVM achieved (86.71%) respectively.

  Comparing One-Class and Two-Class SVM Classifiers for Normal Mammogram Detection

  [Best Student Paper Award].

  Breast cancer is the second leading cause of cancer-related deaths in women in the US. Two main problems appear to affect the decision of detecting and diagnosing breast cancer:the accuracy of the CAD systems used

  and the radiologists’ performance in reading and diagnosing mammograms.In this work we aim to improve CAD system’s performance by adding a preprocessing step to reduce the false negative rate significantly. We propose to divide mammograms into two distinct categories according to tissue type(fatty

  and dense). A one-class classifier is used for each tissue-type separately to enhance the performance of the overall classification task. GLCM features are extracted for each of dense and fatty mammograms. The sensitivity for each tissue type was improved significantly (~100%) when used separately compared to the sensitivity of existing systems (90%) that uses all mammograms regardless of tissue type.

  Detection of Normal Mammograms based on Breast Tissue Density using GLCM Features

  In this work

  we studied the dense mammograms as a distinct category apart from fatty mammograms. One of the factors that affect CAD systems' performance is breast density. The sensitivity of any CAD system will reduce as the density of the breast increases. We enhanced the performance of detection and helped overcome the pitfalls of breast density by separating the mammograms into two distinct categories according to density and performing feature extraction on each tissue type category using tissue-specific features. Our classifier identifies normal mammograms within the same tissue density (dense or fatty). Choosing tissue-specific features for each type of mammogram density will increase the separability between normal and abnormal features and

  therefore

  improve the classification task.

  Screening-out Normal Mammograms using Breast Density Information

  Breast cancer is the second leading cause of cancer deaths in women in the U.S. Two main problems appear to affect the decision of detecting and diagnosing breast cancer: the accuracy of the CAD systems used

  and the radiologists' performance in reading mammograms. The main challenge in designing any CAD system is to maintain a high sensitivity level in detecting the abnormalities as the density of the breast increases. In our work

  we introduce a novel idea of having a dual system that will process mammograms differently according to breast tissue density. The sensitivity will be significantly improved while keeping the specificity as high as possible. Mammograms are divided into two distinct categories according to breast density(fatty

  and dense). Two main set of features are extracted from both dense and fatty mammograms. A one-class classifier is used for each tissue-density separately to enhance the performance of the overall classification task. Results showed that for each density a specific set of features will perform better than others.

  Effect of Breast Density in Selecting Features for Normal Mammogram Detection

  Cost and environmental concerns continue to drive research in high performance computing (HPC) energy optimization. Commodity server platforms are increasingly deployed as compute clusters which have a variety of energy management control features. In this paper

  we examine the energy reduction effect of different ways to co-schedule benchmark codes on a HPC cluster using different combinations of job queue control dimensions including

  thread core affinity interleaving

  Dynamic Voltage and Frequency Scaling (DVFS)

  and job re-ordering. The combination space of control parameters in conjunction with varying job queue depths is too large to explore using a direct measurement approach so we developed a scheduling simulator that can quickly and efficiently examine a large permutation space of job-spans to find the energy optimal order and control configuration. Equipped with the base time/energy profiles of the benchmark algorithms

  the simulator can reliably predict the execution time and energy of all the job queue permutation (ordering) choices

  including the optimal control parameter combinations within a 3% margin of error.

  Energy Efficient Job Co-Scheduling for High-Performance Parallel Computing Clusters

  Breast cancer is the second leading cause of cancer deaths in women in the U.S. Two main problems appear to affect the decision of detecting and diagnosing breast cancer: the accuracy of the CAD systems used

  and the radiologists' performance in reading mammograms. We aim here to improve CAD system's performance by adding a preprocessing step based on the density of the breast to reduce the false negative rate significantly. Mammograms are divided into two distinct categories according to breast density (fatty

  and dense). Three LBP-based features are extracted for each of dense and fatty mammograms. A one-class classifier is used for each tissue-type separately to enhance the performance of the overall classification task. The sensitivity for each tissue type was improved significantly when used separately compared to the sensitivity of existing systems that uses all mammograms regardless of tissue type.

  Pre-CAD System for Normal Mammogram Detection using Local Binary Pattern Features

  New Mexico Consortium

  Los Alamos National Laboratory

  New Mexico State University

  University of Maryland

  New Mexico Consortium