Program Schedule: Regular Poster Sessions
Wednesday, October 1, 2008
7:00 — 9:00 p.m.
SRC Poster Sessions
60GHZ High Speed Wireless Architecture
Candy Yiu (Portland State University)
A wireless architecture for Gbps per user data rate is presented for indoor environments using 60GHz ISM band. We show that antenna geometry has a significant impact on throughput, namely, by utilizing realistic 3D beam patterns. Dead spots and areas with link outages (due to human activity) are covered using novel passive reflectors. Further, mobile users are provided with continuous link connectivity at the cost of lower data rates.
A Case for Adapting Channel Width in Wireless Networks
Ramya Raghavendra (University of California, Santa Barbara)
We study a fundamental yet under-explored facet in wireless communication width of the spectrum over which transmitters spread their signals, or the channel width. Through detailed measurements, we first quantify the impact of channel width on throughput, range, and power consumption. Our findings make a strong case for wireless systems that adapt channel width. We present a channel width adaptation algorithm called SampleWidth, which increases throughput by more than 60% compared to the best fixed-width configuration.
A Novel Authentication and Validation Mechanism for Attack Identification and Trace Back in Syslogs for Forensic Analysis
Steena D.S. Monteiro (Utah State University)
This research proposes a novel technique for authenticating and validating syslogs for forensic analysis. Digital evidence needs to be admissible, authentic, believable, and reliable. The proposed technique uses a modification of the Needham-Schroeder protocol and ties together all the entities involved in the generation of a syslog entry through the assignment of digital fingerprints. The assigned fingerprints and the incorporated challenge response mechanism forensically validate the generated syslogs.
Body Interference and Interconnect Time on Opportunistic Body Sensor Networks
Tammara Massey (University of California, Los Angeles)
The impact of body interference on short range radios, such as 802.15.4, and its implications on capacity management and communication has not been adequately addressed. In this paper, we present a measurement study that isolates and quantifies the effect of body area interference on 802.15.4 radio communication. The connectivity model is combined with mobility data of people to create an algorithm that dynamically adapts to the environment.
Building Robust Intrusion Detection Systems for MANETs Using Cross-layer Designs
Geethapriya Thamilarasu (Univeristy at Buffalo)
In this work, we address the problem of security attacks and vulnerabilities in mobile ad hoc networks. We study the limitations of existing security solutions based on a layered protocol architecture. We introduce novel cross-layer interactions and develop intrusion detection systems (CIDS) to guarantee ad hoc network security. Using simulation results, we demonstrate that our proposed CIDS system enables multi-layer security for MANETs and detects intrusions in the network with improved detection accuracy.
Compatibility-Dependent Application Placement for Energy-Efficient Data Centers
Manal Houri (Southern Methodist University)
Recently, it has been reported that datacenters are responsible for 1.2 to 2.0 percent of the electricity consumed in the United States. Our approach exploits prior knowledge of workload patterns of applications to achieve a power-aware allocation. We propose allocating applications on machines according to the incoming workload characteristics while 1) minimizing the system power consumption, 2) increasing utilization level of running machines, and 3) minimizing the number of running machines.
Computers Can Assist in Retrieving and Assessing New Information in the Medical Domain
Oana Frunza (University of Ottawa)
Systematic reviews are highly structured summaries of existing research in any given field and particularly in the medical domain were they are a fundamental tool for evidence-based medicine. The amount of information in medical publications repositories continues to increase tremendously. Systematic reviews help to parse this growing body of information in order to determine if an abstract/article is relevant to a topic of interest (e.g., cancer, obesity, drugs, treatments etc.).
Crafting Girls into Computing
Kady Rosier (Georgia Institute of Technology)
The Institute for Computing Education (ICE) at Georgia Tech is trying to increase the percentage of students, especially girls, that are interested in computing. We do this by running weekend workshops with the Girl Scouts. In these, they explore different avenues of computing through fun and engaging softwares like PicoCricket and Lego Mindstorms. Our goal is for the girls to think, “Hey, computing can be creative and fun!”
Engendering ICT
Diane P. McCarthy (University of Otago /Christchurch Polytechnic Institute of Technology)
Through a mixed method approach and technofeminist poststructuralist framework, my Master of Arts (Education) thesis seeks to make sense of how women students take up, resist and/or subvert discourses in masculinized IT training programmes. Critiquing current policies and practices may open up transformative strategies in polytechnics and institutes of technology (ITPs), retaining women in ICT by blending feminist practice with the business case to build a better world.
Examination of Variance in Production of Task Operators Questions GOMS Ability for a Quick Semi-Automated Usability Assessment of Interfaces
Maria Vicente Bonto-Kane (North Carolina State University)
GOMS modeling techniques have been exemplary in automating usability evaluation of human-computer interfaces by relying on statistical means for keyboard operators and generating a measure of task complexity and duration. GOMS estimates, however, may be off and its assessment on usability limited. This research considers the need for examining the statistical variance in the production of keyboard operators and the need for better methods for normative standardization of GOMS operators.
HADES - Utilizing Artificial Death for Fault Tolerance
Megan Olsen (University of Massachusetts Amherst)
HADES is a multi-agent system whose agents acknowledge their citizenship to the system and monitor their environment. When an agent senses persistent malfunctioning agents in its neighborhood, it sends anonymous messages to entice them to self-kill. We develop protocols for sending and interpreting these self-kill messages to optimize the system’s longevity. We contribute to AI by introducing protocols for system robustness via notions of active citizenship and programmed death.
HoverCross: Modeless Editing for Pen-Based Computing
Alice (Xuexin) Zhu (Harvey Mudd College)
A central problem in pen-based interfaces is how to transition smoothly between drawing and editing. Separating the two modes can be awkward and distracting, while modeless editing gestures are error-prone. We propose to use the hover space to perform selection by crossing over desired objects in the hover range, providing a simple and reliable method to edit without explicit mode changes. Our user studies indicate that HoverCross generally creates a more fluid transition between drawing and editing.
Impact Analysis of System and Network Attacks using Artificial Neural Networks and Attack Graphs
Anupama Biswas (Utah State University)
Impact Analysis of System/Network Attacks is one of the key areas in security that has not yet been explored extensively. I present the idea of how to predict the impact of a new attack since it is difficult to generalize such an attack while it is in progress. Artificial Neural Networks and Attack graphs have been incorporated to perform the mapping of the novel attack with the existing attacks.
Improving Verification Scalability through Automated Compositional Verification
Haiqiong Yao (University of South Florida)
Formal verification has emerged as a promising complement to simulation for functional verification providing almost exhaustive verification coverage. Its main problem is the state space explosion problem resulting in poor scalability. We develop a framework combining compositional verification and abstraction to address the challenge, and apply it to verify large scale concurrent designs successfully. We develop a novel technique to generate small yet accurate environments automatically to support compositional verification.
Modeling Global Green House Gas Emissions in Proportion to Continental Output
Anastasia Sagalovitch (Baruch College- CUNY)
This project aims to visually model carbon dioxide (CO2) emissions in proportion to amount emitted per continent using POV-Ray, a ray tracing software. This technique was used in order to highlight the critical problem of climate change and the need to reduce green house gas emissions worldwide. GHG emissions data is an integral part of the debate on controlling climate change, thus representing data well is as important as measuring it accurately.
More than Paradoxes to Offer: Exploring Motivations to Attract Women to Computing
Jill Dimond (Georgia Institute of Technology)
We interviewed twelve women at Georgia Tech who are not studying computing and found half had pro-social goals for their careers. They also perceived people who worked in computing to not align with those goals. We showed pictures where technology helped people. Most were really interested in these pictures and given opportunity thought they would be interested in participating - they just never thought of computing as helping people.
Multicasting in Disconnected and Delay Tolerant Networks for Developing Countries
Grace Metri (Wayne State University)
Many around the world live below the poverty line. The only way to improve their life is by education. The best way to spread education is through the internet. Due to the geographic nature of many developing countries, it is difficult and expensive to have reliable networks. Therefore, we explored a way to multicast data to servers across developing countries while adapting current technology to the physical nature of their network.
Physically-Based Granular Dynamics
Tina Yee (University of Toronto)
Granular dynamics refers to the dynamics of a large set of small particles. Convincing simulation remains a challenging computational problem. Our approach begins with a particle-based method to model high velocity interactions. We then add dissipative forces to produce static properties of grains. With these static attributes, we try to characterize the grains and transition to a fluid-based model to reduce computational costs.
Private Searching for Nearest Neighbors
Yinian Qi (Purdue University)
We give efficient protocols for private k-nearest neighbor (k-NN) search, when data is distributed between parties who want to cooperatively compute the answers without revealing their private data. Our single-step k-NN search protocol is secure with linear computation and communication complexity. We also give the general multi-step k-NN search protocol for data with complex distance functions. Our protocols can also be used for k-NN classification and outlier detection.
Quality of Information-Aware Design and Management of Sensor Network
Sadaf Zahedi (University of California, Los Angeles)
Provided quality of information (QoI) by sensor networks to users is an important design goal. Our proposed modular analysis framework evaluates QoI of sensor network deployments. The process is decomposed to steps of modeling the characteristics of sensor networks like faulty behaviors, analyzing QoI at sensor, cluster and network level, and exploring the trade-offs and optimized designs. For reliable designs, sensor network is managed at runtime to detect the faults in tiered fashion and take run-time actions.
Reactions to the Use of Wearable Recording Technology for Aiding People with Memory Impairments
Gabriela Marcu (University of California, Irvine)
SenseCam is a wearable recording device which captures pictures automatically, and was designed to help individuals with memory impairments. Secondary stakeholders’ responses to imagined uses of SenseCam in situ enabled us to construct models about this recording technology grounded in real-life experiences. Our results indicate that self-presentation, control of data and its dissemination, and desire to help those in need all impact reactions to ubiquitous recording technologies like SenseCam.
Reasoning about Multisets with Cardinality Constraints
Ruzica Piskac (Ecole Polytechnique Federale de Lausanne)
Collections are ubiquitous in computer science and reasoning about them is very important in software analysis and verification. There are numerous applications in software analysis that involve reasoning about collections of objects. Cardinality constraints on such collections also arise in these scenarios. Multisets (bags) are introduced by abstracting the content of linked data structure with duplicate elements. We investigate the decidability and complexity of reasoning about multisets with cardinality constraints.
Rich Services: The Integration Piece of the SOA Puzzle
Vina Ermagan (University of California, San Diego)
A major challenge in the development of ultra large scale software intensive systems is the controlled integration of multiple subsystems, such that the resulting system fulfills a wide spectrum of integration requirements, such as authentication, security and governance. To address this challenge, we introduce the notion of Rich Service, an extension of the standard service notion, based on an architectural pattern that allows hierarchical decomposition of system architecture according to separate concerns.
Scaling Network Games Using Collaborative P2P Overlay
Neha Singh (IIT Bombay)
Networked Virtual Environments (NVEs) are computer generated, synthetic worlds that allow simultaneous interactions of multiple participants. Online games representing the most common class of NVEs, are becoming the next frontier for social interaction between humans. A very important aspect in creating rich playing experience is creation of dynamic and complex artificial intelligence (AI) of non-player characters (NPCs) which is computationally expensive. We propose the architecture and techniques to scale computation intensive NVEs using the collaborative peer-to-peer overlay.
2D Shape Decomposition Based on Combined Boundary-Skeleton Features
Jingting Zeng (Temple University)
This project is to implement 2D shape decomposition. Motivated by recent studies in visual human perception discussing the importance of certain shape boundary features as well as features of the shape area, we combine properties of skeleton and boundary to implement an approach leading to shape decomposition. Experiments prove the robustness of our combined approach compared to other decomposition methods.
Tracking in a Spaghetti Bowl: Monitoring Transactions Using Footprints
Animashree Anandkumar (Cornell University)
The problem of tracking end-to-end service-level transactions in the absence of instrumentation support is considered. The transactions progress through a state-transition model and generate time-stamped footprints at each model state. The goal is to track individual transactions using footprints they generate when these footprints may not contain any tokens uniquely identifying the transaction instance that generated them. For a semi-Markov process model, the maximum likelihood tracking rule is explored.
Wednesday, October 1, 2008
7:00 — 9:00 p.m.
Regular Poster Sessions
A Federated Feature for Query-by-Content Audio Retrieval
Yi Yu (Nara Women’s University)
Audio feature extraction plays an important role in query-by-content audio retrieval. In this work we propose a weighting scheme using regression on a database to summarize long audio feature sequences and generate a concise and semantic feature called Federated Feature (FF). Two similarity searching schemes are adopted to test the proposed approach of audio feature summarization. The experimental results demonstrate that our weighting scheme is useful for music content retrieval.
A Goal Oriented Approach for Modeling and Analyzing Security Trade-Offs
Golnaz Elahi (University of Toronto)
In designing software systems, security is only one design objective among many. It may compete with or complement other objectives such as privacy and usability. Often, security mechanisms are adopted without explicit recognition of competing design objectives and their origins in stakeholder interests. We examine how conceptual modeling techniques provide explicit and systematic support for analyzing security trade-offs. We proposed an extension to the i* framework for security trade-off analysis.
A Platform for Developing and Executing Scientific Workflows in Distributed Environments
Corina Stratan (Politehnica University of Bucharest)
In designing software systems, security is only one design objective among many. It may compete with or complement other objectives such as privacy and usability. Often, security mechanisms are adopted without explicit recognition of competing design objectives and their origins in stakeholder interests. We examine how conceptual modeling techniques provide explicit and systematic support for analyzing security trade-offs. We proposed an extension to the i* framework for security trade-off analysis.
A Privacy-Preserving Approach for Spam Filtering Evaluation
Mona Mojdeh (University of Waterloo)
These privacy considerations are often at odds with the conduct of large-scale realistic spam filtering efforts. We address the problem of spam filter evaluation on a realistic email set without the need to sacrifice the privacy at any level. Using our framework, researchers will be able to test different spam filters on real email boxes, for which they will receive the result file (containing no private data) in email.
Accelerating Quantum Monte Carlo Applications on Advanced Computing Platforms
Akila Gothandaraman (University of Tennessee, Knoxville)
We are currently exploring the use of emerging computing platforms in computational chemistry. We present the results when a Quantum Monte Carlo application, which is used to determine the ground-state properties of atomic or molecular clusters, is ported on high performance reconfigurable computing systems. We are also porting our application to platforms such as graphics processing units and multi-core processors and evaluating the performance of our application on these platforms.
Accelerating the Drug Discovery Process with Blue Gene through Massively Parallel Computation and Data Integration
Amanda Peters (IBM)
The pre-clinical phases of drug development consist of the assessment of compound libraries by screening and then by analysis of both biological properties and commercial viability. Developing a solution to accelerate identification and development of drug candidates based on both binding affinities and disease-related receptor knowledge is a key goal for in this field. This work focuses on the methods of leveraging Blue Gene for the integration of massively parallel computation and textual annotation for virtual screening.
Accessibility Assessment Via Workspace Estimation
Jing Yang (York University)
The process of evaluating a built environment for accessibility is known as “accessibility assessment”. Determining accessibility is closely related to the problem of determining possible motions of a specific kinematic structure. Here the accessibility assessment process is reformulated as a motion planning problem. Rather than treating each of the degrees of freedom (DOFs) ’equally’ while planning, we explore a hierarchical characteristic of all the DOFs when constructing the roadmap.
An Energy-aware Relay Selection Scheme for ALLIANCES
Xinhua Yang (Colorado School of Mines)
ALLIANCES is a newly proposed wireless MAC protocol, which exploits the cooperation of source nodes and relay nodes to resolve collisions and further improve throughput. We propose an energy-conserving model, and an Energy-aware Relay Selection scheme (ERS), which maintains the benefits of Location Relay Selection scheme (LRS), but more evenly distributes the energy throughout the network. Simulation results show that the energy-conserving model can save up to 80% energy with the same delay.
An eNotebook Prototype for Students and their Notetaking Tasks
Melissa Bubnash (The University of Montana)
Traditional approaches to managing, storing, and editing electronic data need re-evaluation for today’s college student. This research focuses on the development of a prototype application that allows users to gather and organize files of multiple types, browse the Web, and take notes within a single, seamless user interface. Our goal is to provide students with a better set of tools and centralized data repository for their evolving Information Assimilation tasks.
Antler: A Multi-Tiered Approach to Automated Wireless Network Management
Ramya Raghavendra (UCSB)
Troubleshooting problems related to wireless access in these networks requires a comprehensive set of metrics and monitoring data. Current solutions gather large amounts of data and require significant bandwidth and processing to offload and analyze this management traffic, consequently not scalable or real-time. We propose a multi-tiered approach to wireless network monitoring that dynamically controls the granularity of data collection based on network activity. Our approach can achieve significant bandwidth savings and enable real-time automated management of wireless networks. Our initial analysis using traces from a large WLAN shows significant reduction in amount of data collected to diagnose wireless problems.
BayView Fab Displays: A Collaborative Project for an Automated Manufacturing Facility
Kathleen M. Milhaven (IBM)
The Fab Display or “BayView” project was designed to deliver relevant content and production metrics to computer monitors over a network of computer displays at predetermined locations within the IBM East Fishkill 300mm Manufacturing areas. The focus of this project is on three key areas: 1. Collaborative Customization 2. The “Pushing” of Data, rather than the “Pulling” of Data 3. Presenting Only Critical Data.
BIOCRUISE: An Interactive Graphical User Interface (GUI) to Modeller9v1 for Protein Homology Modeling
Zartasha Mustansar (Comsats University)
BioCruise (BC), an integrated application designed to provide homology modeling under a uniform & user-friendly graphical interface to Modeller library serves to combine the most frequent modeling steps in a semi-automatic, interactive way. BC provides an integrated GUI to the different phases in Homology Modeling such as: protein alignment, modeling, mutation & evaluation. BC is an open-source, cross-platform & extensible framework and allows developers to integrate additional third-party programs.
Bringing stories to the world: The technical challenges of CEOHP
Vicki L. Almstrum (The University of Texas at Austin)
This poster explores technical challenges encountered during the first few years of the Computing Educators Oral History Project (CEOHP). Among the challenges are audio and visual processing, organization, web technology, and ethical/legal issues. We discuss these technical challenges in the context of the processes involved, including data collection of audio interviews and processing to create a transcript and identifying pullquotes, and uses of the processed and approved interviews.
Building a Better World by Growing the Pipeline: The Influence of Collaborative Learning
Cynthia Lester (Tuskegee University)
New data show that students’ interest in computer science (CS) is declining. While there are many suggested reasons for the decline, some educators suggest that the problem is with the way that CS is being introduced to a “technically savvy generation”. In an effort to increase interest in the major, especially among female students, a technique using collaborative learning was infused into an introductory programming course.
Changes in Girls’ Attitudes to Information Technology: A Focus Group Study
Sandra E. Downes (Murdoch University)
There has been a shift in the use of technology by females. The computer is no longer a mathematical device that is only of interest to males. Females are using computers for communication and design and claiming this technology as their own. Using focus groups, this study aimed to investigate this change in perception of IT and whether this could be translated into more females considering a career in IT.
Classified Price Policy and Adaptive Timing for Digital Rights Management
Lingyan Wang (Auburn University)
The popularity of digitalized products and service attracts illegal copying and consumption. Digital Rights Management (DRM), merges to fight against these piracy activities. In this paper, we propose an innovative price policy and timing method as a component of DRM. Particularly, we classify the consumption of a digital product into several levels, each of which requires different price. Then, the expiration of the service is adaptively calculated to prevent illegal use.
Compiler Directed Hierarchical Hardware-Software Mapping for Algorithm-Architecture Codesign
Yuanrui Zhang (Penn State University)
Algorithm-architecture co-exploration and hardware-software codesign have been used to accelerate domain specific applications. We combine these two into a single framework to achieve application performance improvement. The proposed complier directed hierarchical mapping employs an intermediate code representation to capture the program structure, and applies a branch and bound search to find the optimal solution using fast cost estimation models. Experimental results demonstrate that our approach is efficient in finding the optimal mappings for parallel applications.
Complexities and Importance of Successful Cabling Management from Start to Finish
Rebeccah J Vossberg (IBM), Nadia Anguiano-Wehde (IBM)
Cable management is crucial to rack system design and data center layout. It must take into account airflow, cable density, signal integrity, serviceability, and aesthetics in order to produce a successful product. This poster exhibits various hardware, methodologies, and best practices of cable management for high-density racks, as well as Do and Don’ts of cabling for various scenarios.
Computer Science Recruitment for the 21st Century
Jasmine L Glaese (Missouri University of Science and Technology)
This project is creating software to help reverse the trend of decreasing interest in Computer Science (CS) among students, particularly females. Our software is aimed at elementary school students, with an emphasis on female appeal, showcasing the social relevancy of CS through a series of highly visual games & puzzles, illustrating the careers of CS alumni. Survey results show the effectiveness of our software in increasing the students? interest in CS.
Concurrency Management in Project Managed Environments
Johnissia R Stevenson (IBM/University of Arizona)
Concurrency management includes strategy, people/relationships, processes/operations, tools, projects, management, and technology. It integrates and produces results that meet both business and customer needs and expectations. All process and communication channels must interlock to solve problems and improve business. This paper highlights an array of topics on concurrent processes by looking at multiple factors to understand technical project delivery challenges and identify solutions for overall concurrent improvement.
Coping Strategies for CS Women
Valerie Morganson (Old Dominion University)
Coping is a lever for increasing the success and retention of CS women. This poster examines a large sample of CS undergraduates to identify effective coping strategies. Findings indicate that women are inclined to rely upon social resources to cope. Problem-focused coping is related to lower stress, and greater commitment and satisfaction with major for men, but not for women. We provide research-based suggestions for CS students, mentors and teachers.
Could We Improve the Accuracy of Neural Network Derived Optical Coefficients with Boosting?
Jinyan Guan (Winona State University)
To improve the learning performance of Artificial Neural Network on the retrieval of water optical properties from satellite images, we apply boosting to the training samples. The boosting algorithm allows the training process to focus on the ?harder? cases of the training samples. The result shows the boosting can reduce the error rate of absorption coefficient from 9% to 7%.
Crisis Web Portal: Making Sense of On-line Disaster Activity
Amanda Hughes (University of Colorado at Boulder)
Social networking and peer communication on the Internet is generating information that could be used in disaster response and relief efforts. However, the large size and distributed nature of these activities makes it tedious and time consuming to gather and extract. We propose a Crisis Web Portal that automates the process of gathering information sharing activity on the Internet, processing it, and presenting it for use during times of disaster.
Design Strategy for Knowledge Base Formation to Automate a Course Map Creation
Susan Lukose (University of Mississippi)
As popularity of e-learning has increased, the task of assisting an author in creating an online course map or a student in customizing a learning path has a lot of potential. The goal of this research is to create a knowledge base for a decision support system that could guide an author or a student in their respective tasks. We propose to achieve this by automatically creating a ranked ontology from domain specific glossaries.
Developing Student Courses for Wireless Sensor Networks using TinyOS 2.x
Leena Kowser Ganguli (University of Technology, Sydney)
The project, ìWireless Sensor Networks (WSN)î focuses on researching and evaluating methodologies for teaching students how to program on small Networked Sensor Devices known as Motes using TinyOS 2.x. Due to the specialised nature of this topic, this information is not widespread however, WSNs are attracting attention by researchers and commercial entities who foresee the importance of these networks in such areas as health monitoring for the elderly and infirm.
Digging in the Uncertainty: Probabilistic Ranking Queries on Data with Uncertain Linkage
Ming Hua (Simon Fraser University)
Uncertain data is inherent in a few important applications such as environmental surveillance and mobile object tracking. On the other hand, ranking queries are often natural and useful in data analysis. We study probabilistic ranking queries on uncertain data and present efficient query answering algorithms. An extensive empirical study using both real and synthetic data sets are reported.
Digital Library and Corpus of the Collected Works of Ivan Franko
Oresta Tymchyshyn (Lviv Politechnic University)
The poster is devoted to creation of digital library (DL) and Corpus of 50-volume Collected Works of prominent Ukrainian writer Ivan Franko. The technology of preparation of digital publications in both DjVu and PDF formats and principles of presentation of text corpus in the XCES format are described. The method for detecting OCR errors on the basis of Levenshtein algorithm, hidden Markov model and frequency dictionaries is proposed.
Discovering RNA motifs using Generative Models
Hilal Kosucu (University of Toronto)
RNA-binding proteins (RBP) regulate gene expression by binding to target RNAs and thus affecting their translation. Our aim is to characterize the sequence and structure preferences of RBPs using motif models that are fit to semi-quantitative microarray-based measurements of RBP binding affinity to particular RNA sequences. In the present study, we can correctly identify the sequential and structural features of several motifs that are recognized by the RBP VTS1.
ECG based Biometrics
Anupriya Mittal (MIT, Manipal, India)
The electrocardiogram (ECG also called EKG) trace expresses cardiac features that are unique to an individual. The ECG varies from person to person. In fact this is the only biometric token which doesnít exist if the owner is not alive. For ECG processing data filters were designed. Fiducial points were identified on the filtered data and extracted digitally for each heartbeat. From the fiducial points, stable features were computed that characterize the uniqueness of an individual.
Efficient Online Co-Allocation Scheduling Algorithm for Large-Scale Distributed Environments
Claris Castillo (North Carolina State University/IBM)
Complex high performance distributed applications frequently require the co-allocation of multiple resources. In principle, this can be achieved by allocating resources sequentially. However, such a solution can be computationally expensive for time-sensitive applications. We present an efficient online algorithm for co-allocating resources that also provides support for advance reservations. We perform a comparative analysis of our online algorithm against conventional batch schedulers. Our findings indicate that our algorithm may achieve higher utilization while providing smaller delays and better QoS guarantees; and, without adding much complexity. Furthermore, we show that the algorithm proposed scales to systems with large number of resources and heavy workloads.
EXAM - A Comprehensive Environment for Access Control Policy Analysis and Management
Prathima Rao (Purdue University)
As distributed collaborative applications are adopting policy-based solutions for security tasks, management and consolidation of a large number of policies is becoming a crucial component of such solutions. We present EXAM, a comprehensive environment for access control policy analysis and management, which can be used to perform various functions such as policy property analyses, policy similarity analysis and integration. Our work focuses on analysis of policies written in XACML.
Exposing the Cool Side of Building Technology to the Next Generation
Deirdre Athaide (IBM), Jennifer Schachter (IBM)
From Webkins to Pleo to YouTube, the list of incredibly cool, pop culture applications of technology is never-ending. And yet we are faced with the very certain reality that the number of entrants into computer science and other technology related disciplines is declining, particularly for women. Come and join us as we discuss techniques and technologies that can be used to engage middle school aged girls in computer science.
Feedback Controlled Load Balancing for Hybrid Peer-to-Peer Systems
Purvi Shah (University of Houston)
In our prior work we proposed a content delivery network distributing software packages in a large organization. Our design combined the concept of volunteering with the P2P technology. In this work we present a novel load balancing mechanism that considers both the synchronization and the customer-generated workload of the volunteers. Our results indicate that the feedback information currently available at the server/tracker of the P2P system offers enough information to ensure fair load distribution among the peers.
For Better or For Worse: Is What We Are Doing Right?
Jennifer Wong (University of Victoria)
Many initiatives have risen in response to an enrolment decrease in Computer Science. However, many questions and challenges still remain unresolved. Initiatives have been designing and deploying activities and pedagogy, but how do we know we are actually recruiting and retaining instead of dejecting students? This poster will address various issues and attempt to answer some of these issues and challenges.
Formal Computational Models Give Guidance to Initial Design Efforts for Usability
Maria Vicente Bonto-Kane (North Carolina State University)
Decades of work in HCI have looked to using live user testing and GOMS techniques to guide usability design. But with these methods, guidance came after the fact. This research shows how formal computational models can give systematic guidance to initial design. Using formal methods, the probabilistic occurrence of task operators were mapped and optimum placement of frequently accessed icon operators done to ensure design of a more facilitative interface.
GATOR and SWAMP: GPU Computing for Sequence Alignment
Shannon I Steinfadt (Kent State University)
In the search to understand the function of DNA and RNA, a sequence is aligned against one or more other sequences. Smith-Waterman is an exact local sequence alignment algorithm. An existing associative parallel Smith-Waterman algorithm known as SWAMP is in development for two NVIDIA Tesla boards that are general-purpose graphical processing units (GPGPUs). This poster highlights the GPGPU version of SWAMP, or Genetic Alignment on sTream processoRs (GATOR).
Hardware Optimization and Virtualization in Large Datacenters
Clea Zolotow (Regis University/IBM)
Server hardware spending has remained consistent in the last few years. However, power and cooling costs have increased dramatically. This poster shows an overview of p, x, and zSeries virtualization and architecture and provides a high-level methodology for both physical and virtual migration. This allows full utilization of hardware, decreased number of frames, and corresponding power and cooling savings. Case studies and lessons learned are presented.
Helping Bring Technology, Training, and Passion to Health Care in Zambia
Lynn Langit (Microsoft), Chrys Thorsen (TBH Consulting), Esther Schorr (Microsoft)
This poster is based on volunteer work done by technical professionals for a non-profit, healthcare project in Africa. The specific project that the group works on will be used as context for a broader discussion on the ‘why and how’ of matching particular project needs to technical volunteer skills from an international group.
Helping the New Kid on the Block: Recommending Groups in Social Networks
Isabelle Stanton (University of Virginia)
Modern social networks have millions of groups; far more than a user can reasonably be expected to peruse. To deal with this problem we develop a recommendation algorithm for groups in social networks. We base our algorithm on explicitly discovered communities. The primary technical achievement is finding communities in massive networks. Experiments show that exploiting the structure of the network with no text significantly improves the quality of the recommendations.
HospitalLine - A Spoken Dialogue System for Hospitals
Sunayana Sitaram (NIT Surat)
Hospitals in India are staffed with many specialist doctors, each with his/her own schedule for visiting the hospital. Boards at the hospital carry information about their availability but getting information is difficult because they are outdated and use technical terms. Telephone- based enquiry systems are an attractive solution. HospitalLine is a spoken dialogue system, created using the RavenClaw/Olympus framework, using which a user can call the hospital and book an appointment.
How to Combat Plagiarism in Academia (and How Not To)
Jessica Dickinson Goodman (Carnegie Mellon University)
Everyone has heard about students cheating as a ìRip, Mix, Burnî approach to homework. Professors and Administrators struggle to preserve the reputation of their institutions as well as their students’ Intellectual Property. How does one combat plagiarism with so many sources for an unscrupulous student to draw from? This poster will evaluate the current state of both technical and non-technical solutions to academic plagiarism and produce a conclusion on the overall costs and benefits of several methods.
Improving the Performance of Electronic Communities by Sharing Reputation Ratings of Users
Georgia Kastidou (University of Waterloo)
The utility and popularity of self-interested electronic communities in which the provided services are mainly offered by participants are sometimes compromised by the existence of malicious users and free riders. In order to overcome this problem we are investigating ways to motivate agents to increase their contributions by providing incentives to communities to exchange information regarding their agents and to make more sophisticated decisions about which agents to accept.
Incorporating Semantic Information into Text Segmentation
Meghana U Marathe (University of Toronto)
Text segmentation involves the division of a text into cohesive units. Many state-of-the-art segmentation techniques use simple term repetition to measure lexical cohesion. This is mainly due to a lack of reliable semantic information sources. Lexical chains join semantically-related words in a text. Their presence, absence and number help indicate segment breakpoints. Mohammad and Hirst (2006) proposed a framework for measuring concept distance, which we use for chaining. Such chains should outperform term-repetition methods at text segmentation.
Instructional Materials for Engaging Middle School Students in Computing Using Animation
Iris Jomellee Beltran (Lamar University), Valerie Juarez (Lamar University)
It is recognized that there is an increasing demand for computer scientists in todayís market, but university enrollments for computing degrees have been declining in recent years. Studies have shown that studentsí interests are developed by high school, so outreach should target middle school students. This poster presents instructional materials developed to teach concepts of sequence, selection, and repetition in animation using Scratch, a program developed by MIT.
Intelligent Adaptive Scaffolds for Expert Users
Alana Cordick (University of Guelph)
Experts may use a new application due to workplace or higher education requirements. The support required to learn the new software is dependent on the userís demonstrated expertise. Adaptive scaffolds can help these experts to learn the software at a faster rate, using real tasks, increasing productiveness. The adaptive scaffolds allow for the tailoring of support to the expertís need based on the user model.
Learning Styles: How Do They Fluctuate?
Cheryl Swanier (Auburn University), Juan Gilbert (Auburn University)
This article discusses learning styles and how learning styles can vary from lesson to lesson within a specific course. The purpose of this research was to gain insight and hopefully improve teaching methods to facilitate student achievement and retention in the Science, Technology, Engineering, and Mathematics discipline. Through this research, professors will probably be able to answer questions as to why certain students do not perform well on various assessments.
Mining Influential Customers from Online Social Network for Targeted Marketing
Yu Zhang (Zhejiang University)
This paper proposes a method for identifying influential users, based on mining of online social networks. We represent social networks as a directed graph, which incorporates “Web of Trust’’ and “Review-Rate Network’’ on Epinions, and moreover, has a weight associated with each edge to represent the influence of one user on another. We then test several algorithms, including general greedy, hill-climbing and centrality-based algorithms, on the real-world social network to identify influential users.
Modeling Information Roles for Understanding Stock Market Dynamics
Munmun De Choudhury (Arizona State University)
We have developed a framework to determine information roles of people in blogosphere and how their communication is useful to understand external events. This can improvise search algorithms, help determine opinions and understand community evolution. We use a technology-blog Engadget and determine information roles of bloggers which are associated with stock movements of companies. Favorable events are characterized by presence of chatter; while unfavorable events reckon substantial crowd as aftermath.
Multitier Multiscale Sensing: A New Paradigm for Actuated Sensing
Diane M Budzik (University of California, Los Angeles)
Multitier Multiscale Sensing is a new paradigm for actuated sensing for efficiently sampling dynamic spatiotemporal phenomena with high •delity. This approach introduces a hierarchy of sensors according to sampling •delity, spatial coverage, and mobility characteristics. The application of solar light radiation illustrates a two-tier implementation of multiscale sensing. Experiments performed in simulation and on a physical robotic system show that multitier multiscale sensing is suitable for sampling dynamic spatiotemporal phenomena.
Non-Blocking Array-Based Algorithms for Stack and Queue
Niloufar Shafiei (York University)
We present new non-blocking array-based distributed implementations for stack and queue. We give detailed proofs of correctness and amortized time analyses for the algorithms. We use the Spin model checker to verify the correctness of our algorithms. Our second stack and queue algorithms are the first ones that use bounded counter values and our stack implementations are the first practical array-based stack implementations. We compare our stack and queue implementations to the popular proposed implementations experimentally.
On Monitoring the top-k Unsafe Places
Ling Hu (Northeastern University)
In a city, protecting units like police cars move around and protect places such as banks and residential buildings. Different places may under different protection levels by police units. We study the Continuous Top-k Unsafe Places query, which continuously monitors the k least safe places. It is a novel addition to the family of continuous location-based queries. This paper proposes two solutions to this new query and experiments are conducted to evaluate the proposed solutions.
Onto’CoPE: Ontology for Communities of Practice of E-learning
Akila Sarirete (Effat College)
In the present work we focus on the problem of capitalization of techno-pedagogic knowledge, both tacit and explicit in the domain of e-learning. We attempt to solve this problem within the framework of Communities of Practice of E-learning. An ontology will be proposed to model the members of the CoPE,their roles, the learning situations, and annotating the CoPE?s knowledge resources of activities and interactions and the environment of the CoPE.
Protein Representation for Efficient Comparison of Surface Properties
Sael Lee (Purdue University)
Protein surface properties such as shape and physico-chemical properties convey information important in characterizing individual proteins as well as in measuring the affinity of interaction or binding of molecules. Thus efficient representation and analysis of protein surface properties is needed. We proposed the uses of (global/local) 3-deminsional Zernike descriptor to describe the protein surface properties for its compactness and rotation invariance as well as its accuracy.
Representing Process Variation
Borislava I. Simidchieva (University of Massachusetts Amherst)
Processes involving the coordination of human and computer agents to achieve a goal are ubiquitous. Developing and analyzing process definitions can help to improve real-world processes. However, defining real-world processes that exhibit variations is hard. Variations can be modeled with several related definitions, or process variants, which together comprise a family. We introduce an approach for modeling process families using the Little-JIL process definition language, and present a case study.
Scholarship Day
Emily G Raymond (Purdue University)
Scholarship Day is a program of Purdue University’s Computer Science department. The program helps potential students and parents decide if computer science is the best field for them and familiarizes the top scholarship applicants with the Computer Science department. Therefore, the prospective students and their parents get to interact with currents students, faculty, and advisors. Scholarship Day is a unique experience to recruit and retain women to computer science.
Shape Correspondence Using Particle Filter
Shusha Li (Temple University)
The project is to analyze similarities in two shapes. The crucial step is to find the correspondences between shape features in accordance with human perception. The project works on 2-dimensional shapes represented by a polygonal boundary. It utilizes a multiple hypotheses technique, ìParticle Filtersî (PF). Based on certain world knowledge, PF build correct correspondences by evaluating different hypotheses of most reasonable solutions.
Stochastic Analysis of Reversible Self-Assembly
Urmi Majumder (Duke University)
Since Winfree demonstrated that his abstract irreversible self-assembly model is capable of universal computation, self-assembly has been recognized as a promising route towards nano- scale computation. In reality, however, self-assembly is not irreversible, requiring us to develop a reversible model that allows tile dissociation. However, such analysis exists only for one- dimensional assemblies. We extend this work to higher dimensions and describe how self- assembly can be modeled as rapidly mixing Markov Chains.
Study of Pitch Shifts for Query-by-Example Music Retrieval
Yi Yu (Nara Womenís University)
In this poster, a wide set of features covering both spectral and temporal properties are studied. The analysis shows Pitch shift has much less effect on cepstrum than on STFT. Short-Time Fourier Transform (STFT), Mel-Frequency Cepstral Coefficient (MFCC) and two new features based on cepstrum are compared with respect to different pitch shifts. Simulation results verify that retrieval ratio can be improved by removing pitch shift and time variant energy.
Subsumption Architecture and Experiment on team based AI
Siddartha Gautam Janga (University of Memphis), Rajitha Rani Satharla (University of Memphis)
Subsumption architecture for implementation of game AI breaks the system into layers of activities each of which can be designed and debugged independently. Activities are much simpler processes to handle when compared to the overall system they fit into. Designed game AI handles the decision taking process of Bot of first person shooter game. Decisions involve actions such as taking cover, aiming searching for enemies, and team based tactics.
The Correction of GAMER for Multiple Geocasts
Nur Aini Rakhmawati (National Taiwan University of Science and Technology)
Normally, there is not only one source node sending packets to one geocast region in real mobile Ad Hoc environments. We implements GAMER protocol in which there are four source nodes and four different geocast regions. In order to make GAMER feasible for multiple geocasts, we have made some corrections this protocol. Our simulation result shows that the GAMER correction could perform well in multiple geocast regions while its delivery ratio has dropped to nearly 50%.
The Impact of Web Page Usability Guideline Implementation on Aesthetic and E-Retailer Evaluations
Anshu Agarwal (Salesforce.com/Cornell University)
This research examined the impact of front-end usability guideline implementations on web page aesthetics and e-retailer perceptions. A guidelines framework was developed from an in- depth review of the literature. Four design factors (background color, white space, thumbnail image location, and thumbnail image size) were selected and varied using this framework. Sixteen web page prototypes were developed and assessed through use of an online survey. Results provided evidence that subtle design manipulations had significant effects on consumer perceptions.
The Wind of Flocks for Clustering and Data Visualization
Esin Saka (University of Louisville)
The way birds fly is a source of inspiration for challenging problems such as data visualization and clustering. This project presents a bio-inspired algorithm that uses the metaphor of flocks of agents for data clustering and visualization. Our approach makes improvements to overcome some limitations of an existing flock clustering algorithm, known as FClust, and decreases the complexity from quadratic to linear by hybridization with the classical K-means clustering algorithm. Applications to real-life problems are also presented.
Toward Secured Folksonomies
Maryam Ramezani (Depaul University)
Social tagging systems provide an open platform for users to share and annotate their resources such as photos and URLs. Due to their open nature, however, these systems present a security problem. Malicious users may try to distort the systemís behavior by inserting false user profiles. This poster addresses the problem of security and robustness of social tagging systems. We outline a framework to model the navigation of a tagging system. Using our framework we classify attack models and identify different kinds of potential attacks to a social tagging system.
URSA: Understanding User Reviewing Patterns
Gayatree Ganu (Rutgers, The State University of New Jersey)
Websites commonly allow users to input reviews on various topics. These are used when customers buy products and businesses track user feedback. Most reviews are in a free-text format, which is difficult to automatically analyze and aggregate. The User Review Structure Analysis (URSA) project proposes tools to better access and search text-based reviews, augmenting them with structural information, hence improving simple searches and similarity searches in a social network setting.
Using Parallel Computing to Search for High Rank Elliptic Curves
Shweta Gupte (Purdue University)
An elliptic curve is a certain type of cubic polynomial equation. The “rank” of such a curve is a measure of the number of rational points. This project seeks to find curves with “large” rank by sieving through hundreds of millions of examples. The mathematical theory demands, for each example, one search for points on thousands of related quartic curves. For the computing application we use a high performance computing cluster and distribute the search load.
Using Your Brain for Human-Computer Interaction
Erin Treacy Solovey (Tufts)
To increase the bandwidth from humans to computers, we are investigating methods for sensing signals that users naturally give off while using a computer. We then use this data to augment the explicit input that the user provides through standard devices. Using a relatively new brain imaging tool, we can detect signals within the brain that indicate various cognitive states and build interfaces that adapt carefully and appropriately to changes in the user’s cognitive state.
Variations of Strassen’s Matrix Multiplication Algorithms
Sarah M. Loos (Indiana University)
SThe quadrant additions in the classical O(n^lg7), Strassen matrix multiplication algorithms are not ideal for preserving boundaries in rectangular multiplications. Adding the northwest quadrant, which is the only necessarily non-empty quadrant, creates dense sums that, as factors, make more work for six of the seven recursive multiplications. A very simple modification of the traditional algorithms, however, has only three recursions on factors built from that dense northwest block. Experimental results show how the algorithms accelerate by annihilating the other four recursive products, since they are far more likely to have zero factors.
Web Search From a Bus
Aruna Balasubramanian (University of Massachusetts, Amherst)
Intermittent connectivity to wireless Internet access points (APs) from vehicles is commonplace today, but it is inadequate for popular interactive applications such as Web search and browsing. Our work is driven by the questions: How can web search be adapted to tolerate disruptions in connectivity from vehicles? We present Thedu, an architecture that enables interactive Web applications from vehicles. The key idea is to use aggressive prefetching to transform an interactive application into a one-shot request/response process.
Wikipedia as an Ontology for Describing Documents
Zareen S Syed (University of Maryland Baltimore County)
We have investigated the use of Wikipedia as a topic ontology for identifying topics and concepts associated with a set of documents. We have proposed different methods to predict the concepts common to a set of documents. Our experiments show that the Wikipedia category graph can be used to predict generalized concepts, whereas the article links graph helps by predicting specific concepts.
Workplace Culture that Hinders and Assists the Career Development of Women in Information Technology
Rose Mary Wentling (University of Illinois)
This study examined the roles that workplace culture plays in the career development of women in IT. The IT culture has been described as largely white, male dominated, anti-social, individualistic, and competitive. Although these workplace culture characteristics were supported by this study, it was the collaborative and teamwork oriented aspect of their workplace environment and building close relationships with colleagues that benefited the study participants the most in their career development.
ZigBee Enabled Device Location through Trilateration
Kelly A Torkelson (Winona State University)
In wireless networks, it’s beneficial to be able to determine the position of a node using other nodes’ locations. This has been done before, but not yet for ZigBee. We will show how we can estimate the location of a node in a ZigBee network using received signal strengths. The methodology we will use is trilateration, which uses three devices to estimate the position in a two dimensional plane.