Department of Computer Science and Electrical Engineering

Tyler Campbell, Nick Ducq, Ryan King, Andrew Nguyen and Tim Spillman walked out of the Baltimore Convention Center elated. Their team, the Sherwood Cyber Warriors, had just won the high school division of the inaugural Maryland Cyber Challenge. Their success netted them each a $5,000 scholarship from the National Security Agency.

The entire experience was rewarding for both the students and their parents, says Steve Weiss, one of the team’s advisers. “Winning first place was the icing on the cake.”

In a conference with over 800 attendees, the excitement over the cyber competition was palpable. With scoreboards changing in real time, onlookers crowded around to see who was in the lead.

Following the competition, held October 21 and 22, eight teams from each division — professional, college and high school — walked away with scholarships and cash prizes. The scholarships for students, put up by the National Security Agency (NSA), totaled more than $84,000.

Members of first place high school and college teams took home $5,000 scholarships each. Members of second place high school and college teams took home $2,000 scholarships. Each member of the first place professional teams won $2,000 and each second place member won $1,000.

First place winners in the college and high school categories were from, respectively, the University of Maryland, College Park and Sherwood High School. Second place winners were Towson University and Poolesville High School. In the professional category Team ICF came in first, with Team Pr3tty coming in second.

The Sherwood Cyber Warriors, four seniors and a junior, are mostly undecided in their future careers, although one does plan to work cybersecurity. Jim Kirk, the team’s senior advisor, says that regardless of the their ultimate career choices, the students learned valuable skills from the competition — such as how to communicate effectively and work as a team.

The Cyber Warriors began practicing for the competition in May, often meeting twice a week. The team developed strategies to pick the low hanging fruit — what hackers go for first. That, says Kirk, includes developing strong passwords and removing unnecessary software from servers.

The challenge for the high school teams, says Rick Forno, Director of UMBC's Graduate Cybersecurity Program and an organizer of the Cyber Challenge, was purely defensive. “They were being attacked and their job was to keep services open.”

The challenge was run using CyberNEXS, a software system developed by SAIC for cybersecurity training and exercises.The system is self-contained and runs both Windows and UNIX systems.

But, more than just the chance to compete, the event gave college and high school students a taste of what cybersecurity work is like. And that, involves more than technical skills say professionals in the field.

“The cyber challenge is especially interesting to me, since all the students participating are passionate about cyber security and the teams will only excel if every member is doing their job,” says Neil Furukawa, vice president of CyberPoint International. “We’re looking for people who can lead, but who can also roll up their sleeves and get the work done.”

Phyllis Villani, Director of Talent acquisition at Northrop Grumman says that to get a job, “networking is key.” Besides honing “soft skills” like communication, Furukawa says, people should never stop their education because cybersecurity is a rapidly evolving field.

Fittingly, education is what the Maryland Cyber Challenge is all about.

Originally posted by Nicole Ruediger at November 18, 2011 1:02 PM

MS Thesis Defense

Analyzing the Growth of Hoeffding Trees

Mayank Sawhney
12:00-1:30pm Thursday 1 December 2011, ITE 346

Mining high speed data streams has become a necessity because of the enormous growth in the volume of electronic data. In the past decade, researchers have suggested various models for learning in both stationary and concept drifting data streams. Hoeffding Trees (Domingos & Hulten 2000) are one such model for mining stationary data streams. Several modifications of the nave Hoeffding Tree algorithm have been proposed to study data streams.

Our work analyzes the behavior of Hoeffding Trees when they are trained on infinite and experiments, we show that the Hoeffding bound suffers from an inherent shortcoming. Even after reaching a stage where accuracy asymptotes, Hoeffding Trees continue to grow. We examine this behavior in data streams with both nominal and numeric attributes. We also study enhancements made to the naive Hoeffding Tree algorithm and also evaluate different discretization methods.

In our work, we analyze how the Hoeffding bound relates to the information gain when splits are made and also when we send a random distribution as a data stream. We conclude that this behavior is a result of decisions made for the early growth of Hoeffding Trees and the induced randomness in an online setting. We also argue that the presence of this behavior will impact the use of Hoeffding algorithms in real world online applications.

Committee Members

• Dr. Tim Oates (Chair)
• Dr. Tim Finin
• Dr. Kostas Kalpakis

Masters Thesis Defense

Approximation of Nonintegral Frequency Moments

Brian Pilz

10:00am 30 November 2011, ITE325b

Let a data stream have length m over an alphabet of n letters, with letter i occurring m_i times for i = 1,…,n. For any k, define the frequency moments F_k as F_k = sum_{i=1}^n m_i^k. Alon, Matias, and Szegedy showed how to estimate F_k for integers k>0 with a one-pass algorithm using O(n^{1-1/k}log n) space for given length m, accuracy, and confidence. Here we extend those results to non-integral k obtaining bounds on the variance giving accuracy and confidence estimates, and giving quantitative results on the algorithm’s space requirements with particular interest to when k is near 1. We also give some performance statistics of the algorithm for these cases and consider an application to entropy estimation. This algorithm is known as a sketching algorithm. Sketching algorithms are probabilistic algorithms generally requiring sublinear space vs. a "classical" O(n) (linear) space requirement, and may have applications for anomaly detection of systems or networks.

Committee:

• Drs. Samuel Lomonaco
• Brooke Stephens
• Kostas Kalpakis (chair)
• Larry Wagoner

EE Graduate Seminar

Time-Varying Amplitude Compression Processing to
Preserve and Enhance Spectral Contrast in Speech Signals

Dr. Janet C. Rutledge
Dean, UMBC Graduate School
Vice-Provost for Graduate Education
Affiliate Associate Professor of Electrical Engineering

11:30-12:45 Friday, 18 November 2011, ITE 231

Multichannel amplitude compression processing is used to reduce the level variations of speech to fit the reduced dynamic ranges of listeners with sensorineural hearing loss. This processing, however, can result in smearing of temporal information, artifacts due to spectral discontinuities at fixed channel edges, and spectral flattening due to reduced peak-to-valley ratios. Presented here is an implementation of a time-varying compression processing algorithm based on a sinusoidal speech model. The algorithm operates on a time-varying, stimulus-dependent basis to adjust to the speech variations and the listeners hearing profile. The algorithm provides fast-acting compression with minimal artifact, has time-varying frequency channels, is computationally inexpensive and preserves the important spectral peaks in speech.

This method has been extended to provide real-time enhancement of spectral peaks and valleys. This work is also related to processing audio signals that will be transmitted over amplitude-limited noisy channels or for listeners in a noisy environment.

Dr. Janet Rutledge is Dean of the Graduate School and Affiliate Associate Professor in the CSEE Department at UMBC. She received the BS in electrical engineering from Rensselaer Polytechnic Institute and the MS and Ph.D. in electrical engineering from Georgia Tech. Prior to coming to UMBC in 2001, she was a faculty member at Northwestern University, and program director at the National Science Foundation.

Host: Prof. Joel M. Morris

Ph.D. Dissertation Defense

Identifying and Isolating Text Classification Signals
from Domain and Genre Noise for Sentiment Analysis

Justin Martineau

1:30-4:00 Friday, 18 November 2011, ITE 325b, UMBC

Sentiment analysis is the automatic detection and measurement of sentiment in text segments by machines. This thesis provides methods to identify, characterize, and isolate the sentiment bearing terms to improve textual sentiment classification when there is little or no labeled data for the domain.

We introduce a new theoretical framework that explains the different sources of noise that affect term level sentiment bias. This noise comes from the genre the author communicates in and the domain or general topic that the author is writing about. To understand the affects of domain noise we defined sentimental domain independence and statistically described it in the multi-domain product review data set. This allowed us to design a Domain Independence Verification Algorithm (DIVA) to eliminate this noise and produce a domain-independent sentiment model using data drawn from a variety of different domains. This model is the most accurate method to classify documents in the 25 category product review data set.

Committee:

• Dr. Tim Finin (chair)
• Dr. Marie desJardins
• Dr. Akshay Java
• Dr. James Mayfield
• Dr. Tim Oates

The Department wishes to extend its congratulations to the Computer Science and Electrical Engineering graduate students that were recently admitted to candidacy for the doctoral degree. The following students were recognized at the Doctoral Candidates Recognition Reception that was held last Tuesday, November 1st in the UC Ballroom:

Computer Science

David Chapman
Mentor: Dr. Milton Halem
"Multidimensional Map Reduce for Remote Sensing Applicators

Niyati Chhaya
Mentor: Dr. Tim Oates
"Joint Inference for Extracting Text Descriptors from Triage Images of Mass Disaster Victims"

Yasaman Haghpanah
Mentor: Dr. Marie desJardins
"A Trust Model for Decision Making in Supply Chain Management"

Lushan Han
Mentor: Dr. Tim Finin
"GoRelations: An Intuitive Query System for Linked Data"

Karuna Joshi
Mentor: Dr. Yelena Yesha, Dr. Tim Finin
"Framework for an Integrated Lifecycle of Virtualized Services"

Varish Mulwad
Mentor: Dr. Tim Finin
"Generating Linked Data by Inferring the Semantics of Tables"

Randy Schauer
Mentor: Dr. Anupam Joshi
"A Framework for the Intelligent Management of Distributed Rack-Based Blade Operations"

Fatih Senel
Mentor: Dr. Mohamed Younis
"Relay Node Placement for Federating Segmented Wireless Sensor Network"

Shiming Yang
Mentor: Dr. Konstantin Kalpakis
"Improving the Traffic Flow Forecasts for Road Networks with Data Assimilation"

Xianshu Zhu
Mentor: Dr. Tim Oates
"Finding Story Chain in Newswire Articles"

Electrical Engineering

Matthew Anderson
Mentor: Dr. Tulay Adali
"Joint Blind Source Separation with Multivariate Gaussian Model: Algorithms, Performance Analysis and Applications"

Damon Bradley
Mentor: Dr. Joel Morris
"On the Analysis, Modeling, and Mitigation of Radio Frequency Interference for Spaceborne Microwave Radiometers"

Sai Ma
Mentor: Dr. Tulay Adali
"Analysis of Brain Network Connectivity in MRI Data Using Spatial Information"

Ganesh Saiprasad
Dr. Chein-I Chang
"Adrenal Gland Abnormality Detection Using Random Classification Forest"
 

EE Graduate Seminar

Functional Brain Circuits, nAChR Genes, and Smoking

Dr. Elliot Hong
Director, Brain Imaging Center
Associate Professor
Maryland Psychiatric Research Center
University of Maryland School of Medicine

11:30am-12:45pm Friday, 4 Nov 11, ITE 231

About 20% of the US population smokes cigarettes. Smoking is influenced by genetic and environmental factors and mental illnesses.The neurobiological basis of severe nicotine addiction is unclear. We use gene-circuit analysis, resting fMRI, diffusion tensor imaging, and circuit-addiction behavior analyses to examine the dorsal anterior cingulate and the ventral striatum/extended amygdala (dACC-VS/EA) brain circuit and its relationship to smoking.

Dr. Hong received the M.D. degree in 1986 from the Sun Yat-Sen University Science in China. In 2003-07 he was an Asst Professor in the Maryland Psychiatric Research Center (MPRC), Psychiatry Department, University of Maryland School of Medicine. He became an Assoc. Professor in 2008. In 2010 Dr. Hong was appointed as both the Chief of the Neuroimaging Research Program and Director of the Brain Imaging Center at MPRC.

Seminar Host: Prof. Joel M. Morris

DARPA has announce another research-relevant competition: the $50,000 Shredder Challenge which invites teams to try to reconstruct virtual shredded documents.

"Today’s troops often confiscate the remnants of destroyed documents in war zones, but reconstructing them is a daunting task. DARPA’s Shredder Challenge calls upon computer scientists, puzzle enthusiasts and anyone else who likes solving complex problems to compete for up to $50,000 by piecing together a series of shredded documents.

The goal is to identify and assess potential capabilities that could be used by our warfighters operating in war zones, but might also create vulnerabilities to sensitive information that is protected through our own shredding practices throughout the U.S. national security community. …

The Shredder Challenge is comprised of five separate puzzles in which the number of documents, the document subject matter and the method of shredding will be varied to present challenges of increasing difficulty. To complete each problem, participants must provide the answer to a puzzle embedded in the content of the reconstructed document."

You can download the puzzles, register a team and submit solutions online as well as view the current top teams. The prizewinner and prize awarded will depend on the number and difficulty of the problems solved. DARPA will a winner in the week of December 5, 2011 once final results are calculated.

CSEE Colloquium

Identifying and Isolating Text Classification Signals from
Domain and Genre Noise for Sentiment Analysis

Justin Martineau
Computer Science and Electrical Engineering
University of Maryland, Baltimore County

1:00pm Friday 4 November 2011, ITE 227

Justin Martineau will describe the results of his PhD dissertation which he will defend later this month. His dissertation research makes both algorithmic and theoretical contributions to the fields of domain adaption and sentiment analysis. First, it provides algorithms to discover and weight discriminative classification task specific features within a domain. Second, it produces algorithms to score how well these features transfer to a new target domain. Third, it lays out a general theory for the kinds of information and the types of noise they produce that exist in text classification tasks. Finally, the dissertation presents a definition of domain independence and a statistical description of it. The research offers readers a firm theoretical foundation as well as practical algorithms when implementing any of the motivating examples and for future research in the field.

As part of Google’s ongoing commitment to encouraging students of underrepresented backgrounds in technology to pursue graduate study, the company will host the 2012 Google Graduate Researchers in Academia of Diverse backgrounds (GRAD) CS Forum. This forum will bring together students who are historically underrepresented in the field to connect with one another and with Google.

Up to 75 computer scientists will be invited to an all-expenses paid forum that will run Wednesday evening through Friday afternoon on January 18–20 at Google’s offices in Mountain View, CA and San Francisco, CA.

The Google GRAD CS Forum will include technical talks from established researchers – both from Google and universities – and a unique occasion to build and strengthen networks with other emerging researchers. Students will also enjoy tours of the Googleplex, have the opportunity to meet with Google engineers in their focus areas, and have fun exploring the San Francisco Bay Area.

Eligibility Requirements. Applicants must

• be a computer science (or related technical discipline) graduate student currently enrolled in a Masters or PhD program at a university in North America
• demonstrate academic excellence and leadership in the computing field maintain a cumulative GPA of at least 3.3 on a 4.0 scale or 4.3 on a 5.0 scale or equivalent in their current program

How To Apply. Applicants will be asked to provide:

• a current copy of your resume
• unofficial or official copies of your transcripts from both your undergraduate and graduate degree-granting institutions
• a brief thesis abstract or description of your current research (500 words or less)

Please note that recommendation letters are not required.

The forum is open to all qualified graduate students, and is committed to addressing diversity in our company and in the technology industry. Students who are a member of a group that is historically under-represented in the technology industry are encouraged to apply, including women, Native American, African American and Hispanic students as well as students with disabilities. Please send any questions directly to . They look forward to reviewing your applications! Apply online by Sunday, November 13, 2011 at 11:59 p.m. PST.