Link to
Colorado State University Home Page

BMAC Spring 2008: Abstracts


Network Coding: A New Paradigm For Networking
Dr. Anthony Ephremides
Cynthia Kim Professor of Information Technology, University of Maryland


Abstract
Network Coding is a relatively recent idea that has the potential to revolutionize the foundations of Networking. Both the Internet architecture as well as that of wireless networks may have to be thoroughly redesigned if the idea of Network Coding proves to be implementable. In this talk we will review and explain the principles of Network Coding, highlight the major results, identify its shortcomings, and show how the basic idea can be adapted for wireless networks. We will selectively describe some performance and design issues in more detail but we will aim at a high-level presentation accessible to a broad audience.

BIO

Anthony Ephremides received his B.S. degree from the National Technical University of Athens (1967), and M.S. (1969) and Ph.D. (1971) degrees from Princeton University, all in Electrical Engineering. He has been at the University of Maryland since 1971, and currently holds a joint appointment as Professor in the Electrical Engineering Department and the Institute of Systems Research (ISR). He is co-founder of the NASA Center for Commercial Development of Space on Hybrid and Satellite Communications Networks established in 1991 at Maryland as an off-shoot of the ISR. He was a Visiting Professor in 1978 at the National Technical University in Athens, Greece, and in 1979 at the EECS Department of the University of California, Berkeley, and at INRIA, France. During 1985-1986 he was on leave at MIT and ETH in Zurich, Switzerland. He was the General Chairman of the 1986 IEEE Conference on Decision and Control in Athens, Greece. He has also been the Director of the Fairchild Scholars and Doctoral Fellows Program, an academic and research partnership program in Satellite Communications between Fairchild Industries and the University of Maryland. He won the IEEE Donald E. Fink Prize Paper Award (1992). He has been the President of the Information Theory Society of the IEEE (1987), and served on the Board of the IEEE (1989 and 1990). He is a Fellow of the IEEE.

Dr. Ephremides' interests are in the areas of communication theory, communication systems and networks, queueing systems, signal processing, and satellite communications.


Meeting the Challenges of the 21st Century Earth System Modeling at the Petascale and Beyond
Richard D. Loft
Techhnology Development Division NCAR


Abstract
Dramatic improvements in computing power, along with rapid advances in disk capacity, tape densities and network bandwidths, are transforming how our society generates and consumes information. Earth system science is no exception. The availability of observational data about our planet in digital form, coupled with once-unimaginable computer modeling capabilities, has allowed geoscientists to tackle a broad front of complex, interdisciplinary, grand challenge problems in new and more realistic ways. For example, climate scientists, once confined to low-resolution simulations using atmospheric general circulation models with prescribed sea surface temperatures, now work with fully dynamically coupled models of sea ice, ocean and land surface processes. In the future, additional processes, such as ocean and atmospheric chemistry, dynamic vegetation, and the carbon cycle will be included, and resolution dramatically increased. Trends toward increased interdisciplinarity and complexity are recapitulated across many grand challenge problems in computational geoscience, ranging from space weather to modeling subsurface fluid flow. The feasibility of deploying advanced models to tackle these problems is complicated by two factors: first, the architectural trends of supercomputing systems, which point towards increased levels of parallelism, and second, for such systems to be useful, scientists from many disciplines, distributed across many institutions, will need to share vast amounts of data seamlessly. Realizing this vision of integrated distributed cyberinfrastructure for geoscience research is no simple task. The talk will show the progress made to date, by NCAR and other institutions, to meet these challenges through improvements in application scalability, development of distributed data federation systems, and the creation of national-scale grids for high performance computing such as the TeraGrid. Finally, to accomplish these ambitious goals, the next generation of scientists and engineers must be inspired, educated and trained. Programs and opportunities at NCAR designed to introduce students to applied mathematics, high performance computing, and computational geoscience will be presented.

BIO

Dr. Loft has been involved with parallel computing since joining Thinking Machine Corporation as an Application Engineer for NCAR in 1989. Throughout his career he has contributed to the understanding and effective use of parallelism as applied to grand challenge climate simulations. He parallelized NCAR’s Community Climate Model (CCM-2) using data parallel CM Fortran for the Connection Machine (CM-2 and CM-5) supercomputers. Dr Loft has been involved with application development for Beowulf technology since 1997. He has created an efficient 3-d spectral dynamical core called Built on Beowulf (BOB). BOB out performs CCM-3 by a factor of five on certain tests, and has been used to study jet formation on Jupiter. Dr. Loft’s career has been driven by an interest in the interplay of algorithms, software design and optimization techniques to achieve flexible, high performance modeling capabilities. Dr Loft, along with CSS collaborators, developed an efficient spectral element based primitive equations core on the cubed-sphere. This work was recognized with an honorable mention prize in the IEEE/ACM Gordon Bell competition at Supercomputing 2001. Most recently, Dr. Loft has taken an interest in advancing the capabilities of end-to-end biogeochemistry models such as Biome-BGC and the PCTM models, which was used for the global carbon cycle forecast component at the C-DAS meeting held in May of 2002 at NCAR.


Translating Thoughts into Actions by Finding Patterns in Brainwaves
Chuck Anderson
Colorado State University


Abstract
Weak electrical signals generated by the brain were first observed on the scalp in the early 20th century. By the middle of the 20th century, patterns in brainwaves were found that are associated with movements, some actually preceding the movement. The finding that similar patterns occur with imagined movement has motivated recent efforts to translate these patterns into signals for controlling a wheelchair or a prosthetic arm. This could provide a paralyzed person some control of their environment and restore the ability to communicate for someone who has lost all voluntary muscle control. Recent work has carried the search for brainwave patterns beyond imagined movements to mental tasks, such as multiplication, counting, and music recall. The objective of this line of research is a practical "brain-computer interface", preliminary examples of which will be discussed.

Bio:

He graduated with a Ph.D. in computer science from the University of Massachusetts, Amherst, in 1986, and worked at GTE Laboratories in Waltham, MA, until hr arrived at CSU in 1991. He works with neural networks, reinforcement learning, EEG pattern recognition, neural modeling, HVAC control, adaptive tutoring, computer graphics, computer vision, and software and hardware testing.


Vulnerability Discovery in Multi-Version Software Systems with Shared Source Code effect
Dr. Yashwant K. Malaiya
Colorado State University


Abstract
The vulnerability discovery process for a program describes the rate at which the vulnerabilities are discovered. A model of the discovery process can be used to estimate the number of vulnerabilities likely to be discovered in the near future. Past studies have considered vulnerability discovery only for individual versions, without considering the impact of shared code among successive versions and source code evolution. Here we examine a new approach for quantitatively modeling the vulnerability discovery process, based on shared source code measurements among multi-version software systems. The applicability of the approach is examined using Apache HTTP Web server and Mysql DataBase Management System (DBMS). We examine the relationship between shared code size and shared vulnerabilities between two successive versions. We find that vulnerabilities continue to be found for an older version because part of its code is shared by the newer, and more popular version. Thus even when the installed base of an older version has declined, vulnerabilities applicable to it are still discovered. With the proposed modeling approach for multi-version software, we show that the vulnerability discovery trend is explained well when the shared code is taken into account.

BIO

Yashwant K. Malaiya is a Professor in Computer Science Department at Colorado State University. He has published widely in the areas of security vulnerabilities, fault modeling, software and hardware reliability, testing and testable design. He served as General Chair of IEEE International Symposium on Software Reliability Engineering (ISSRE), Denver, 2003; IEEE Asian Test Symposium (ATS), Shanghai, 1999; General Chair, Sixth International Conference on VLSI Design (VLSI Design '93), Bombay, India, 1993. He has co-edited the IEEECS Tech. Series books ``Software Reliability Models, Theoretical Developments, Evaluation and Applications'' and ``Bridging Faults and IDDQ Testing''. He was a guest editor of special issues of IEEE Software and IEEE Design & Test. He received the IEEE Third Millennium Medal, 2000, and IEEE CS Golden Core Award, June 1996 for services to IEEE Computer Society


Scientific Computing: Applications, Algorithms, Architectures
Dr. Paolo Bientinesi
Duke University


Abstract
Research in the fields of scientific and high-performance computing deals with the development of fast and accurate numerical algorithms. Until recently, a strict separation of concerns has resulted in limitations that favored generality in place of performance and accuracy. A collaborative approach that combines knowledge from applications, algorithms and architectures would result in computations exploiting the structure and properties of the application at hand, as well as the features of the target architectures. Ideally, a computer system would be able to identify and take advantage of these properties and features automatically. In this talk, I present a project that exemplifies the aforementioned synergistic approach; the objective is the fast computation of a stream of Fast Fourier Transforms on the Cell Broadband Engine. The Cell processor is an emerging multi-core low-power processor capable of attaining remarkably high performance on single precision computations. We devised an algorithm for two-dimensional and three-dimensional FFTs that takes full advantage of the architectural features of the Cell. In addition, in sharp contrast with existing approaches, our algorithm is incremental, so that the final result can be visualized (and tested) incrementally, as the computation proceeds. Experimental results show competitive performance.

BIO

Dr. Paolo Bientinesi was born and grew up in Italy, on the coast of Tuscany. In 1998, he received his Laurea degree (equivalent to M.S.) in Computer Science at the University of Pisa. In 2000, he moved to the US to join the graduate school at The University of Texas at Austin in the Department of Computer Sciences. Six years later, during a hot Texan week of July 2006, not only did Italy win the football World Cup, but he also received his Ph.D. under the supervision of Prof. Robert van de Geijn. His dissertation "Mechanical Derivation and Systematic Analysis of Correct Linear Algebra Algorithms", was later selected as the department candidate for the 2006 ACM Doctoral Dissertation Award. He is currently working as research associate in the Department of Computer Science at Duke University, collaborating with Prof. Xiaobai Sun.


Precise Program Analysis with Data Structures
Bor-Yuh Evan Chang
University of California, Berkeley, Computer Science Department

Abstract
Program analysis tools are being adopted by industry to improve the reliability and overall quality of software like never before because they can rule out entire classes of errors. Yet, today's tools are far from being as effective as they could be, for almost all program analyses have difficulty when objects of interest are put into data structures. Program analyses that reason precisely about data structures typically require sophisticated (and thus often burdensome) logical invariant specifications from the user. We propose a novel way to involve the user in guiding the analysis by extracting both the necessary invariants and reasoning rules from executable assertions in the code. In this talk, I describe a new technique for precise program analysis in the presence of data structures. It is based on data structure validation code that is often written anyway for testing purposes. From the developer's perspective, such validation code provides guidance to the analysis in a familiar style, and we show how our analysis results can be rendered graphically in a form that is comparable to what might be drawn on a whiteboard or printed in a textbook. From the analysis tool's perspective, data structure validation code provides the essential ingredients for a good abstraction that precisely represents the important facts while ignoring irrelevant details. The crucial innovations in our system are automatic methods for understanding and generalizing the developer-provided data structure validation code to make them useful for static program analysis. Example results produced by our analysis tool, Xisa, are available at http://xisa.cs.berkeley.edu/.

BIO

Bor-Yuh Evan Chang is completing his Ph.D. with George Necula at the University of California, Berkeley working in the areas of programming languages and program analysis. He is interested in tools and techniques for building, understanding, and ensuring reliable computational systems. His current focus is on using novel ways of interacting with the programmer to design more precise and practical program analyses.


Storage Systems for Global Scale Datacenters
Hakim Weatherspoon
Cornell University

Abstract
Digital information plays an increasingly critical role in scientific research, military systems and other enterprises, and this trend has important implications. First, many systems are more and more being distributed over a global network of datacenters, which is emerging as an important distributed systems paradigm. Second, storage systems in these environments must ensure the durability, integrity, and accessibility of digital data, and do so under potentially turbulent conditions. For example, in large scale distributed systems, servers continuously fail; data should remain durable despite constant failure. Antiquity is a distributed storage system designed for these sorts of challenging environments. It maintains data securely, consistently, and with high availability in a dynamic wide-area environment. At the core of the system is a novel secure log structure that permits Antiquity to guarantee the integrity of stored data, even under extreme stress. Data is replicated on multiple servers in a manner that ensures that it can be retrieved later even when some replicas are inaccessible. Moreover, unlike prior fault-tolerant systems, the Antiquity fault-tolerance protocols can handle high levels of node churn, regenerating data on the fly when necessary to handle faults ranging from server outages to Byzantine (malicious) attacks. Further, I will present SMFS, a remote mirroring solution targeted for settings where high-speed high-latency links connect a pair of datacenters. SMFS provides strong disaster tolerance guarantees with asynchronous performance-mirroring response times are more typical of high-speed LAN setting. Not only does the approach provide reliability through mirroring, but there are conditions under which it offers dramatic power savings. Longer term, we see SMFS and Antiquity as two examples of a family of innovative solutions addressing a range of demanding problems seen in turbulent, mission-critical, and power constrained settings.

BIO

Hakim Weatherspoon is currently a post-doctoral fellow at Cornell University. His work covers various aspects of information systems, distributed systems, network systems, and peer-to-peer systems with focus on fault-tolerance, reliability, security, and performance of Internet-scale systems. He previously received his Ph.D. from University of California, Berkeley in Computer science.


Productive Supercomputing
Dr. Grzegorz Malewicz


Abstract
The success of MapReduce at Google demonstrates the importance of a simple parallel computing model supported by an execution system realizing the model in the presence of load imbalances and failures at the scale of a datacenter. The talk will overview the system.

BIO

Grzegorz Malewicz received the BA degrees in computer science and in applied mathematics in 1996 and 1998, respectively, and the MS degree in computer science in 1998, all from the University of Warsaw. He received the PhD degree in computer science from the University of Connecticut in 2003. He is a senior engineer at Google. He has had internships at the AT&T Shannon Laboratory (summer 2001) and Microsoft Corp. (summer 2000 and fall 2001). He visited the Laboratory for Computer Science, Massachusetts Institute of Technology (MIT, academic year 2002-2003), and was a visiting scientist at the University of Massachusetts, Amherst (summer 2004) and Argonne National Laboratory (summer 2005). He was an assistant professor at the University of Alabama, where he taught computer science from 2003 until 2005. His research focuses on high-performance parallel and distributed computing, experimental and theoretical algorithmics, combinatorial optimization, and scheduling. His research appears in top journals and conferences and includes a singly authored SIAM Journal on Computing paper that solves a decade-old problem in distributed computing.


Data: Making it be there when you want it and go away when you want it gone
Dr. Perlman
Cynthia Kim Professor of Information Technology, University of Maryland


Abstract
In order not to lose data, copies should be kept in lots of locations. That makes it difficult to really delete the data, since the backup copies can be stolen or copied. The obvious solution is to encrypt the data, and then discard the keys of data that is to be destroyed. However, reliably keeping, then reliably destroying all copies of deleted keys has the same problem. This talk describes a system that supports three types of assured delete; expiration time known at file creation, on-demand deletion of individual files, and custom keys for classes of data. It is easy and inexpensive to manage and involves only trivial performance overhead over a traditional encrypted file system.

BIO

Radia Perlman Radia Perlman is a software designer and network engineer sometimes referred to as the 'Mother of the Internet'. She is most famous for her invention of the spanning-tree protocol, while working for Digital Equipment Corporation, which is fundamental to the operation of network bridges. She also made large contributions to many other areas of network design and standardization such as link-state protocols. She obtained a Bachelor's, Master's in Mathematics, and a Ph.D. in Computer Science from MIT. Her doctoral thesis at MIT addressed the issue of routing in the presence of malicious network failures and forms the basis for most of the work in this field. Radia is the author of two textbooks on networking. She is currently employed by Sun Microsystems. She holds more than 47 patents from Sun alone.


Cartesian computations and the high cost of moving data
Dr. Larry Carter
UC San Diago


Abstract
In this talk, we identify and analyze a class of algorithms that includes many familiar and important scientific computations. A "2-D Cartesian computation" is characterized by having two very large data structures, A and B (perhaps A is the input and B the output), and for each suitably chosen chunk of A and chunk of B, there is a chunk of computation that must be performed. When neither A nor B fits in the fast memory of a computer, the time (or energy) needed to move bits between cores, chips, nodes and levels of the memory hierarchy can dominate the computation. Static Partitioning, Tiling, Inspector/Executor strategies, and Bucketizing are some well-known programming techniques that reduce data movement. We present a methodology that, for many Cartesian computations, allows one to decide which is the best of these techniques. Our results elegantly relate three orthogonal aspects of a computer -- computation speed, memory capacity, and communication or memory bandwidth -- and show that different techniques are needed at different levels of architectural granularity.

BIO

Larry Carter received the A.B. degree from Dartmouth College in 1969 and the Ph.D. in mathematics from the University of California at Berkeley in 1974. He worked at as a Research Staff Member and manager at IBM's T.J. Watson Research Center for nearly 20 years in the areas of probabilistic algorithms, compilers, VLSI testing, and high-performance computation. From 1994 to 2004, Dr. Carter was a professor in the Computer Science and Engineering Department of the University of California at San Diego. Between 1996 and 2000, he served as Vice Chair and then Chair of the department. His current research interests include scientific computation, performance programming, parallel computation, and computer architecture. Prof. Carter is a Senior Fellow at the San Diego Supercomputing Center, a Fellow of the IEEE, and a Professor Emeritus at UCSD.


Myths, Missteps, and Folklore of Network Protocol Design
Dr. Perlman
Cynthia Kim Professor of Information Technology, University of Maryland


Abstract
Network protocol design is not a nice, clean science, where what gets deployed is the best possible design. Instead, designs are influenced by issues such as politics, general confusion, and backward compatibility. Statements get made, and repeated, until it never occurs to anyone to question whether they're true. Mistakes get made, and rather than backing up and fixing them, kludges are introduced to make things sort of work. This talk discusses how some of the odder things we live with (e.g., bridges) came about, and interesting bad protocol designs that have been standardized and/or deployed. It also discusses "obvious" protocol design issues that somehow get overlooked, such as designing for future evolution, and ability to change parameters, node by node, without disrupting a network. The talk is intended to be provocative, making people question the things they have always taken for granted. It is also a plea to teach the subject in a way that empowers students to think critically about protocol designs, rather than simply memorizing the current standards in order to implement them.

BIO

Radia Perlman See above.


Quality Now Requires - Small Delay Fault Model Dr. T.W. (Tom) Williams
Synopsys Fellow, Synopsys, Inc.


Abstract
The concept of small delay faults has been discussed for more than 20 years. Methods for determining the relative merits of delay test sets have also been known for 20 years. Until recently this area of testing has been considered unnecessary. Today many groups want to use small delay fault testing to achieve high quality levels. This lecture will address the derivation of quality metrics and how they are used today. Testing for small delay defects requires ATPG-FS tools to understand timing information of the design such that transition delay faults can be detected along longer paths. Timing information is analyzed for use in test automation tools to test for small delay defects. Fundamentals of static timing analysis are analyzed with regard to test. This lecture shows that Signal Integrity information can be ignored by test automation tools when timing information is used to guide ATPG tools towards longer paths. This work also shows that a lack of understanding of clock trees in the long path ATPG algorithm leads to incorrect results.

BIO

Dr. Thomas W. Williams Dr. Thomas W. Williams is a Synopsys Fellow at Synopsys in Boulder, Colorado, U.S.A. Formerly, he was with IBM Microelectronics Division and manager of the VLSI Design for Testability group. He received a B.S.E.E. from Clarkson University, an M.A. in pure mathematics from the State University of New York at Binghamton, and a Ph.D. in electrical engineering from Colorado State University. He has received numerous best paper awards from the IEEE and ACM and was twice a Distinguished Visitor lecturer for the IEEE Computer Society. Dr. Williams has previously served on the Computer Society Board of Governors and the IEEE Board of Directors, and was the Society's 2000 Treasurer. He is a member of the Eta Kappa Nu, Tau Beta Pi, IEEE, ACM, Sigma Xi, and Phi Kappa Phi. He is an Adjunct Professor at the University of Calgary, Calgary, Alberta, Canada. Dr. Williams was named an IEEE Fellow in 1988 and received the Computer Society's W. Wallace McDowell Award for outstanding contributions to the computer art in 1989. He was named a member of the Chinese Academy of Science. In 2007 Dr. Williams received the European Design and Automation Association Lifetime Achievement Award for "outstanding contributions to the state of the art in electronic design, automation, and testing of electronic systems."


CS Research Symposium: Session 1

Generation of Data-Flow Analyses with DFAGen
Andrew Stone, Michelle Strout, Shweta Behere

Abstract: Data-flow analysis is a commonly used technique to gather program information for use in transformations such as register allocation, dead-code elimination, common sub-expression elimination, scheduling, and others. This paper presents a tool, DFAGen, that allows compiler writers to specify, and generate, data-flow analyses using a succinct specification language.

Seasonality in Vulnerability Discovery in Windows Operating Systems
HyunChul Joh and Yashwant Malaiya

Abstract: This study examines whether vulnerability discovery rates exhibit an annual seasonal pattern using the seasonal index and autocorrelation function approaches. A time series analysis that can combine the longer term trends with cycles caused by seasonality may predict the future pattern more accurately. The study examines the data sets for four major Windows operating systems obtained from National Vulnerability Database. The analysis shows that there is indeed an annual seasonal pattern with higher incidence during the middle of winter and summer seasons

Use of A New Trust Model for Making Reasoned Decisions in Different Security Contexts
Sudip Chakraborty and Indrajit Ray

Abstract: Security services rely to a great extent on some notion of trust. However, there is no accepted formalism or technique for the specification of trust and for reasoning about trust. In this paper we present an overview of a new trust model and discuss how this model helps to make reasoned decision in different security contexts. For example, in access control for open and distributed systems or, for finding a 'trusted path' to deliver data from a source to a destination in an ad hoc network.


Optimization of Strategies/Heuristics for Delay Tolerant Ad-Hoc Networks Dr. Pascal Bouvry
Sciences, Technology and Communications of Luxembourg University


Abstract
Delay tolerant mobile ad-hoc networks (DTN) and hybrid networks require new generations of protocols and middleware in order to enable context-aware services and mobile grid computing. The underlying optimization issues are multi-objective by nature: e.g. optimizing the bandwidth use, the cost and efficiency of such services. We propose an approach based on the use of meta-heuristics for fine-tuning parameters of distributed lightweight strategies/heuristics for local decision-making. The fitness function representing the global behavior of the network relies on network characteristics such as network density and mobility models. We demonstrate the use of this approach for broadcasting and information gathering on DTNs, for trust management for MANETs, and for choosing injection points for hybrid networks. New generations of meta-heuristics such as co-evolutionary and cellular genetic algorithms are used for the optimization process.

BIO

Dr. Pascal Bouvry Pascal Bouvry earned his undergraduate degree in Economical & Social Sciences and his Master degree in Computer Science with distinction ('91) from the University of Namur, Belgium. He went on to obtain his Ph.D. degree ('94) in Computer Science with great distinction at the University of Grenoble (INPG), France. His research at the IMAG laboratory focused on mapping and scheduling task graphs onto Distributed Memory Parallel Computers. Next, he performed post-doctoral research on coordination languages and multi-agent evolutionary computing at CWI in Amsterdam, the Netherlands. Dr Bouvry gained industrial experience as manager of the technology consultant team for FICS (NASDAQ: SONE) a world leader in electronic financial services. Next, he worked as CEO and CTO of SDC, a Saigon-based joint venture between SPT (a major telecom operator in Vietnam), Spacebel SA (a Belgian leader in Space, GIS and Healthcare), and IOIT, a public research and training center. After that, Dr. Bouvry moved to Montreal as VP Production of Lat45 and Development Director for MetaSolv Software (NASDAQ: ORCL), a world-leader in Operation Support Systems for the telecom industry (e.g. AT&T, Worldcom, Bell Canada, etc.). Dr. Bouvry is currently heading the Computer Science and Communications (CSC) research unit of the Faculty of Sciences, Technology and Communications of Luxembourg University, and serving as Professor. Pascal Bouvry is also a member of the administration board of CRP-Tudor and a member of various scientific committees and technical workgroups (ERCIM WG, COST TIST, LIASIT, etc.).


Neural Stimulation using Implantable Devices: Auditory and Optical Systems
Dr. Timothy Starkweather
Consultant, Second Sight Medical Products


Abstract
In this presentation, Dr. Starkweather will cover the work he has been doing which enables the deaf and severely hearing impaired to hear and newer work which strives to give blind people rudimentary vision for increased mobility. This presentation will delve into the history of cochlear implants and the subjects that can benefit from this technology. Details of a system of audio input, gain control, filters, and audio to electrical mapping will be presented. This will then be extended to current work which involves stimulation of the optic nerve through an electrode array attached to the retina. The talk will include discussion of the patient fitting system, digital signal processing, telemetry system, and implant operation. For one hour prior to the talk, a demonstration of actual hardware and firmware operation will be presented.

BIO

Timothy Starkweather Dr. Starkweather graduated from Colorado State University in January of 1993. Since that time, he has been involved in software, firmware and system design for a variety of implantable medical devices that include Pacemakers, Implantable Defibrillators, Implantable Insulin Pumps with patient telemetry control combined with long-term implantable glucose sensors, Cochlear Implants and, currently, a visual prosthesis for individuals who have lost their sight in adulthood.

CS Research Symposium: Session 2

A Structured-Outputs Method for Prediction of Protein Structure
Artem Sokolov and Asa Ben-Hur

Abstract: We apply the structured-output methodology to the problem of predicting the molecular function of proteins. Our results demonstrate that learning the structure of the output space yields better performance when compared to the traditional ``transfer of annotation'' method.

A Taxonomy of Capabilities Based DDoS Defense Architectures
Vamsi Kambhampati, Dan Massey, Christos Papadopoulos

Abstract: Distributed Denial of Service (DDoS) attacks pose immense threat to the Internet. In this paper, we explore a new class of DDoS defense architectures called capabilities. These architectures advocate a fundamental change to the Internet architecture, so that senders must obtain permission from a receiver before they are allowed to send traffic. Our work re-examins the existing proposals in capability architectures from ground up, and identifies crucial challenges in building a capabilities enable Internet. To this end, we develop a taxonomy of capability architectures with the intent to better understand the architectural changes, and the engineering tradeoffs to make the paradigm shift in the Internet.

Classifier Bias in Protein Function Prediction
Mark Rogers and Asa Ben-Hur

Abstract: In view of its importance, development of novel methods for protein function prediction is an active area of research in bioinformatics. Many researchers assess the accuracy of their methods without reference to the source of the annotations used for training their models. This can lead to over-optimistic results: Given that many existing annotations are based on sequence or structural similarity, it is no surprise that a classifier that uses such information can predict such annotations with high accuracy. We illustrate this phenomenon in a controlled set of experiments using a simple nearest-neighbor classifier that uses PSI-BLAST similarity scores.