Department of Computer Science and Electrical Engineering

UMBC CSEE Colloquium

Energy Efficient Platforms for
High Performance and Embedded Computing

Dr. Tinoosh Mohsenin
Computer Science and Electrical Engineering
University of Maryland, Baltimore County

1:00pm Friday, 7 December 2012, ITE 227, UMBC

Future embedded, high performance, and cloud computing must meet limited energy capacity, cost, and sustainability. These devices will regularly execute over one tera-operations per second (TOPS) with a variety of diverse workloads—from baseband communications to wearable medical devices—while operating on a 5 to 25 Watt-hour cellphone/tablet battery. The need for greater energy efficiency, smaller size and improved performance of these devices demands a co-optimization of algorithms, architectures, and implementations. This talk presents several programmable and application specific solutions that illustrate the cross-domain optimization.

The design of system-on-Chip blocks becomes increasingly sophisticated with emerging real-time computational and limited power budget requirements. Two such algorithms, Low Density Parity Check (LDPC) decoding and Compressive Sensing (CS), have received significant attention. LDPC decoding is an error correction technique which has shown superior error correction performance and has been adopted by several recent communication standards. Compressive sensing is a revolutionary technique which significantly reduces the amount of data collected during acquisition. While both LDPC decoding and compressive sampling have several advantages, they require high computational intensive algorithms which typically suffer from high power consumption and low clock rates. We present novel algorithms and architectures to address these challenges.

As future systems demand increasing flexibility and performance within a limited power budget, many-core chip architectures have become a promising solution. The design and implementation of a programmable many-core platform containing 64 cores routed in a hierarchical network is presented. For demonstration, Electroencephalogram (EEG) seizure detection and analysis and ultrasound spectral doppler are mapped onto the cores. The seizure detection and analysis takes 900 ns and consumes 240 nJ of energy. Spectral doppler takes 715 ns and consumes 182 nJ of energy. The prototype is implemented in 65 nm CMOS which contains 64 cores, occupies 19.51 mm2 and runs at 1.18 GHz at 1 V.

Dr. Tinoosh Mohsenin is an assistant professor in the Department of Computer Science and Electrical Engineering at the University of Maryland Baltimore County since 2011. Prior to joining UMBC, she was finishing her PhD at the University of California, Davis. Dr. Mohsenin’s research interests lie in the areas of high performance and energy-efficiency in programmable and special purpose processors. She is the director of Energy Efficient High Performance Computing (EEHPC) Lab where she leads projects in architecture, hardware, software tools, and applications for VLSI computation with an emphasis on digital signal processing workloads. She has been consultant to early stage technology companies and currently serves in the Technical Program Committees of the IEEE Biomedical Circuits & Systems Conference (BioCAS), Life Science Systems and Applications Workshop (LiSSA), International Symposium on Quality Electronic Design (ISQED) and IEEE Women in Circuits and Systems (WiCAS).

More information and directions: http://bit.ly/UMBCtalks

Center for Hybrid Multicore Productivity Research (CHMPR)
Distinguished Computational Science Lecture Series

Parallel Real-Time OLAP on Multi-Core Processors

Frank Dehne
Chancellor's Professor of Computer Science
Carleton University, Ottawa, Canada
http://www.dehne.net

3:00 p.m. Thursday, 6 December 2012, ITE 456, UMBC

One of the most powerful and prominent technologies for knowledge discovery in Decision Support systems is On-line Analytical Processing (OLAP). Most of the traditional OLAP research, and most of the commercial systems, follow the static data cube approach proposed by Gray etal. and materialize all or a subset of the cuboids of the data cube in order to ensure adequate query performance. Practitioners have called for some time for a real-time OLAP approach where the OLAP system gets updated instantaneously as new data arrives and always provides an up-to-date data warehouse for the decision support process. However, major problems for real-time OLAP are significant performance issues with large scale data warehouses. The aim of our research is to address these problems through the use of efficient parallel computing methods. We present a parallel real-time OLAP system for multi-core processors. To our knowledge, this is the first real-time OLAP system that has been parallelized and optimized for contemporary multi-core processors, providing the opportunity for real-time OLAP on large scale data warehouses. Our system allows for multiple insert and multiple query operations (transactions) to be executed in parallel and in real-time. We evaluated our method for a multitude of scenarios (different ratios of insert and query transactions, query transactions with different sizes of results, different system loads, etc.), using the TPC-DS “Decision Support” benchmark data set. The tests demonstrate that our parallel system achieves a significant speedup in transaction response time and a significant increase in transaction throughput. Since hardware performance improvements are currently achieved not by faster processors but by increasing the number of processor cores, our new parallel real-time OLAP method has the potential to enable OLAP systems that are real-time and efficient/feasible for large databases.

MS Thesis Defense

Simultaneous Feature Acquisition and Cost Estimation

Zachary Kurtz

11:00am Thursday, 6 December 2012, ITE 325b

This thesis will address classification problems with two sources of cost: the cost of acquiring feature values and the cost of incorrect classifications. In particular, I address problems with feature costs and instance-dependent misclassification costs. Many real-world applications, such as medical diagnosis, contain both feature acquisition costs and instance-dependent misclassification costs. The goal of my research is to minimize the total cost of classifying an unknown instance. This goal is accomplished with a new approach: Simultaneous Feature Acquisition and Cost Estimation (SFACE), which combines feature acquisition methods with a regression algorithm that estimates misclassification costs. The estimated cost values are used to estimate the expected cost reduction for the acquisition of each feature. SFACE is evaluated by comparing the total cost of operation to the cost incurred by existing cost-insensitive, cost-sensitive, and feature acquisition algorithms. The results show that SFACE results in lower total cost for the tested datasets.

Committee: Dr. Marie desJardins (Chair), Dr. Tim Oates and Dr. Michael Grasso

MS Thesis Defense

Stateless Detection of Malicious Traffic:
Emphasis on User Privacy

Paul Halvorsen

1:00pm Monday, 3 December 2012, ITE 346, UMBC

In order to allow flexibility in deployment location and to preserve user privacy we have performed research into stateless classification of network traffic. Stateless detection allows for flexibility in deployment location because traffic on a network does not necessarily follow the same path to and from the end points. By only requiring a single direction of traffic, we have the ability to deploy this classifier anywhere on a network. We also do not require the data from a packet which preserves user privacy and allows for the classification of encrypted traffic.

Our research shows that it is possible to determine if traffic is malicious by using packets traveling in a single direction and without the data contained in the packet. Our research shows that with the use of the timing of the packets, time to live value, and source and destination IP addresses and ports, it is possible to determine if the traffic is malicious. In this way we are able to deploy the classifier anywhere on a network, preserve user privacy, and classify encrypted traffic.

Committee members:

• Dr. Anupam Joshi (chair)
• Dr. Charles Nicholas
• Dr. Tim Finin

ACM Distinguished Speaker

Cloud based Active Archiving Solution for Databases

Dr. Mukesh Mohania
IBM Research – India

2:30pm Friday, 30 November 2012
Room 102 (LH8), ITE Building, UMBC

In the second talk of the UMBC ACM Student Chapter's Tech Talk Series, ACM Distinguished Speaker Dr. Mukesh Mohania will visit UMBC and talk about "Cloud based Active Archiving Solution for Databases".

Cloud computing offers an exciting opportunity to bring on-demand applications to customers and is being used for delivering hosted services over the Internet and/or processing massive amount of data for business intelligence. In this talk, we will discuss the architecture of cloud computing, MapReduce, and Hadoop. We will then discuss how the cloud infrastructure can be used for data management services, how the massive amount of data can be processed over cloud for various business intelligence applications, and how the cloud can be used for 'Active' Data Archival for near real-time data access. We discuss various issues concerning the active archive system including schema modification, query federation, query optimization, access control and data provenance. Using TPC-DS benchmark data, we present evaluation results that shows the ability of our system to seamlessly query archive data along with data stored in the warehouse in order of minutes compared to hours required to move the data into the warehouse from traditional archival systems.

Mukesh Mohania received his Ph.D. in Computer Science & Engineering from Indian Institute of Technology, Bombay, India in 1995. Currently, he is a Senior Technical Staff Member and IBM Master Inventor in IBM Research – India. He has worked extensively in the areas of distributed databases, data warehousing, data integration, and autonomic computing. He has published more than 120 papers and also filed more than 50 patents in these or related areas, and more than 14 have already been granted. He received the best paper awards in CIKM 2004 and CIKM 2005. His work on Data Quality, Information Integration, and Autonomic Computing has led to the development of new products and also influenced several existing IBM products. He has received several awards within IBM, such as "Excellence in People Management", “Outstanding Innovation Award”, "Technical Accomplishment Award", “Leadership By Doing”, and many more. He also received IEEE Meritorious Service Award. He is an ACM Distinguished Scientist, and a member of IBM Academy of Technology.

Light refreshments will be served after the talk outside ITE-325

RSVP via Facebook https://facebook.com/events/378277722253548/

More information and directions: http://bit.ly/UMBCtalks

UMBC CSEE Colloquium

Advanced Computer Systems (ACS)
Machine Learning program

Mark McLean
Senior Researcher, Advanced Computer Systems group

1:00pm Friday, 30 November 2012, ITE 227, UMBC

My talk will discuss the ACS Machine learning program. The ACS ML program's focus is on three main areas; algorithm development, applied research and integration into efficient hardware. Our algorithm development work has created the Concurrent Learning Algorithm and Importance Map technologies. These technologies were developed in-house and have some unique capabilities which make it ideal for our purposes. I'll give some demos of these technologies learning on datasets from the UCI repository. For our current research effort, I will discuss our ideas of using neural networks to process complex digital algorithms, which is not a traditional focus for neural networks. Here, I will discuss our efforts to make a neural network learn the Advanced Encryption Standard encryption functionality and why this could impact the way we design digital systems in the future. For our hardware focus, I'll talk about our efforts to develop a Memristor-based neuromorphic processor and why we hope to succeed where others have failed.

Mark McLean has been a senior researcher in the Advanced Computer Systems group Since 2009. His main area of research is on neural network algorithms, application and neuromorphic processor development. Mr. McLean has done post-graduate work at UMD, Holds a MS degree in Computer Engineering from Loyola College and a BS degree in Computer Science. Previously, he held the position of technical director for the microelectronics and reverse engineering group in the DOD. He has work in industry as lead designer for re-configurable computing at Annapolis Micro-Systems and is a retired officer from the USAF.

— more information and directions —

Information Systems Department Seminar

Subjectivity and Social Role Recognition in Meetings

Theresa A. Wilson
Human Language Technology Center of Excellence
Johns Hopkins University

12 noon-1pm, Tue. 20 Nov. 2012, ITE459

Opinions, sentiments and other types of subjective content are an important part of any meeting. Meeting participants express pros and cons about ideas, they support or oppose decisions, and they make suggestions that may or may not be adopted. In this talk, I will present an annotation scheme for labeling subjective content in meetings, as well as experiments for recognizing subjective utterances and their polarity. Our experiments show that even very shallow linguistic features, such as n-grams of characters, can be effective for this task, and that the combination of classifiers using word, character, and phoneme n-grams yields the best result for subjective utterance recognition. Finally, I will discuss the application of subjectivity recognition to social role recognition in meetings.

Theresa Wilson is a research scientist working on opinion and sentiment analysis at the Johns Hopkins Human Language Technology Center of Excellence (HLTCOE). Before coming to Johns Hopkins, she completed her post-doctoral research as part of the AMIDA Project (www.amiproject.org) at the University of Edinburgh Human Communication Research Centre. She received her Ph.D. in Intelligent Systems from the University of Pittsburgh in 2008.

MS Thesis Defense

Smartphone Application and Data Privacy Control Using Semantically Rich Reasoning and Context Modeling

Dibyajyoti Ghosh

9:00am Tuesday, 20 November 2012, ITE 325B, UMBC

We present our ongoing work on user data and contextual privacy preservation in mobile devices through semantic reasoning. Recent advances in context modeling, tracking and collaborative localization has led to the emergence of a new class of smartphone applications that can access and share embedded sensor data. Unfortunately, this also means significant amount of user context information is now accessible to applications and potentially others, creating serious privacy and security concerns. Mobile OS frameworks like Android lack mechanisms for dynamic privacy control. We show how data flow among applications can be successfully filtered at a much more granular level using semantic web driven technologies that model device location, surroundings, application roles as well as context-dependent information sharing policies.

Committee members:

• Prof. Anupam Joshi (Chair)
• Prof. Tim Finin
• Prof. Yelena Yesha
• Prof. Shujia Zhou

Ph.D. Thesis Defense Announcement

Data Intensive Scientific Compute Model for Multicore Clusters

Phuong Nguyen

10:00am 21 November 2012, ITE 325B

Data intensive computing holds the promise of major scientific breakthroughs and discoveries from the exploration and mining of the massive data sets becoming available to the science community. This expectation has led to tremendous increases in data intensive scientific applications. However, data intensive scientific applications still face severe challenges in accessing, managing and analyzing petabytes of data. In particular, workflow systems to support such scientific applications are not as efficient when dealing with thousands and even more of complex tasks within jobs that operate across high performance large multicore clusters with very large amounts of streaming data. Scheduling, it turns out, is an integral workflow component in the execution often of thousands or more tasks within a data intensive scientific application as well as in managing  the access and flow of many jobs to the available resource environment. Recently, MapReduce systems such as Hadoop, have proven successful for many business data intensive problems. However, there are still many limitations in the use of MapReduce systems for data-intensive scientific problems mainly because they do not support the characteristics of science such as data formats, specialized data analytic tools (e.g. math libraries), accuracies, and interfaces with non MapReduce components.

This thesis addresses some of these limitations by proposing a MapReduce workflow model and its runtime system using Hadoop for orchestrating MapReduce jobs for data intensive scientific workflows. Novel heuristic based scheduling algorithm is proposed in the workflow system to manage the parallel execution of data intensive scientific applications. This thesis has developed a hybrid MapReduce scheduling algorithm based on dynamic priorities, proportional resource sharing techniques that reduce delays for variable length concurrent tasks, and takes advantage of data locality. As a result, a new scheduling policy, Balanced Closer to Finish First (BCFF), is proposed as solutions for some problems of scheduling in MapReduce environment. The scheduling algorithm is implemented in Hadoop 1.0.1 framework and is available as a new Hadoop plug-in Scheduler. The evaluations of the workflow system on the climate data processing and analysis application (several TB dataset) show that it is feasible and significantly improved compared to traditional parallel processing method. The scientific results of the application provide new source of monitoring global climate changes for the near decade 2002-2011.

Thesis Committee:

• Prof. Milton Halem (Chair)
• Prof. Yelena Yesha (Co-Chair)
• Prof. Tim Finin
• Prof. Yaacov Yesha
• Prof. Tarek El-Ghazawi at George Washington University

Ph.D. Thesis Defense Announcement

A Decadal Gridded Hyperspectral Infrared Record for Climate
Sep 1st 2002 – Aug 31st 2012

David Chapman

2:00pm 20 November 2012, ITE 325B

We present a gridded Fundamental Decadal Data Record (FDDR) of Brightness Temperatures (BT) from the NASA EOS Atmospheric Infrared Sounder (AIRS) from ten years of hyperspectral Infrared Radiances onboard the NASA EOS Aqua satellite. We show that these results are consistent with the expected greenhouse forcings, and also discovered a drift of ~0.13K/decade in spectrum relative to 4 MODIS-Aqua for Global All-sky Brightness Temperatures. AIRS, operational on September 1, 2002 is the first successful hyperspectral satellite weather instrument of more than 1 year, as well as the longest running global IR hyperspectral measurement. Although global surface temperature data records are available for over 130 years, it was not until 1978 when the Microwave Sounding Unit (MSU) was the first instrument series to reliably monitor long-term trends of the upper atmosphere. The Atmospheric Infrared Sounder (AIRS) provides the first continuous global hyperspectral IR radiance data record from a single satellite for a decade. Our contribution, was to prepare a gridded data record from the AIRS Outgoing Longwave Spectrum (OLS). We have shown high correlations with the GISS global surface air temperatures as well as with the NOAA ONI index of El Niño phase. In addition, we have performed inter-annual inter-comparisons with the Moderate Resolution Spectroradiometer (MODIS) on the same satellite to examine the relative consistency of their calibrations. The comparisons of the two instruments for the 4µ spectral channels indicate an inter-annual warming of 0.13K per decade of AIRS more than MODIS. This relative decadal drift is small relative to inter-annual variability but on the order of historic surface temperature trends. In the 12µ window channels we see a relative constant difference of 0.01K over a decade. It is convenient to observe the climate variability by using monthly average lat-lon grid projections. The polar orbiting data projection to a lat-lon grid is a lossy process that invariably introduces aliasing artifacts and noise. We observed an exponential decay between the number of days averaged and the expected noise due to gridding. We have extended the Observation Coverage (Obscov) gridding algorithm, developed for the MODIS instrument that incorporates the Point Spread Function (PSF) and we show the Obscov gridding algorithm reduces the aliasing noise from AIRS grids by nearly 40% by comparing the spatial correlation of gridded MODIS IR data. We also show that the use of a circular approximate PSF is a sufficient representation to obtain the noise reduction of Obscov at the climate resolution 0.5×1 degree monthly average grids. We extended these spatial sampling methods to the AIRS Level 3 retrieval records for which quality filtering due to opaque clouds is an additional spatial sampling challenge. We correct for an observed dry bias in the AIRS Level 3 monthly average gridded moisture retrieval records by means of spatial interpolation with the Nearest Neighbor (NN) strategy.

Thesis Committee:

• Dr. Milton Halem  (Chair)
• Dr. Yelena Yesha
• Dr. Chin-I Chang
• Dr. Shujia Zhou
• Dr. Joel Susskind (NASA Goddard)