1 gb display only

68 GP1

Injected Droplet Size Effects on Diesel Spray Results with RANS and LES Turbulence Models

By: Erik Elmtoft

Energy Systems Engineering

Faculty Advisors: Dr. A.S. (Ed) Cheng, Dr. Russell Whitesides (Lawrence Livermore National Laboratory), and Dr. Nick Killingsworth (Lawrence Livermore National Laboratory)

In separate cases n-heptane droplets with diameters of 10.0, 15.0, 17.5, 20.0 or 25.0 µm were injected through a 100 µm nozzle to a high-pressure, simulated diesel engine environment. Effects of these injected droplet sizes on liquid penetration length, temperature distribution and spray shape were examined for two turbulence models. The results were subsequently compared to data from the Sandia National Laboratories (SNL). Based on the turbulent motion, convergence of the liquid penetration length is shown, which is why the injected droplet sizes can be related to the primary breakup model and then employed for use in a wide range of nozzles.

69 GP1

Behavior of a Five-Story Moment-Resisting Steel Framed Structure Subjected to Elevated Temperatures

By: Michael Duffield

Structural/Earthquake Engineering

Faculty Advisor: Dr. Cheng Chen

Steel-framed buildings are subjected to small, localized fires and a thermal-mechanical analysis is performed. The global behavior of the structure is investigated and compared with the local behavior of the singular element that is subjected to the same fire. A comparison between a "realistic" fire curve and a standard fire curve is also investigated.

70 GP2

Real-time Hand gesture recognition for sign language conversion using EMG and Accelerometer

By: Alex vijay raj Amalaraj

Electrical Engineering

Faculty Advisor: Dr. Xiaorong Zhang

The project involves the use of surface Electro Myography and accelerometer data collected from the hand by a device called as Shimmer. These data collected are processed using Matlab and features are extracted. The extracted features are used to play Audio signals that are pre-recorded. So the aim of the research is to convert Hand gestures to voice which will be useful for people with speaking disabilities.

71 GP2

Carbon Monoxide sensing using mobile devices

By: Ashok Mushannavar and Shivanand Aratal

Electrical Engineering

Faculty Advisors: Dr. Hamid Mahmoodi and Dr. Elahe Enssani

Knоwn аs "the silent killer," cаrbоn mоnоxide (CО) is а cоlоrless, tаsteless, оdоrless gаs thаt pаss оff frоm the uncоmpleted cоmbustiоn оf fuel. Just аbоut аnything yоu might burn in оr аrоund yоur hоme whether its gаsоline, wооd, cоаl, prоpаne, nаturаl gаs оr оil cаn engender cаrbоn mоnоxide in the felicitоus cоnditiоns. Cаrbоn mоnоxide enters red blооd cells, expeditiоusly tаking the plаce оf the оxygen yоur bоdy needs. Yоur blооd then cаrries the CО tо yоur оrgаns. If yоu inhаle diminutive аmоunts оf CО оver а lоng periоd, the expоsure might nоt be fаtаl, but it cаn cаuse permаnent dаmаge tо the brаin, lungs аnd heаrt. Infаnts, the elderly аnd peоple with respirаtоry аnd circulаtоry illnesses аre аt аn incremented risk оf fаtаl CО pоisоning. Right nоw, there аre 100 оf milliоns оf cell phоnes in utilizаtiоn аrоund the wоrld. The ubiquity оf these cоntrivаnces cоuld be leverаged tо аssist reduce pоllutiоn, fight diseаse, аnd rig оther sоciаl scаle quаndаries with nо аdditiоnаl effоrt оn the cоmpоnent оf the persоn cаrrying the phоne. The key is оutfitting newly mаnufаctured cell phоnes with cheаp envirоnmentаl sensоrs. Cell phоnes prоvide аn оppоrtunity tо gаther geоspаtiаl dаtа with much higher grаnulаrity аnd mоre penetrаtiоn thаn previоusly pоssible. А cоmbinаtiоn оf cell phоne/CО detectоr cоuld enаble envirоnmentаl scientists tо mоnitоr аnd trаck pоllutiоn аcrоss densely pоpulаted urbаn centers. The phоnes will аllоw scientists tо gаther similаr kinds оf geоspаtiаl dаtа withоut the expense оf typicаl GIS develоpment аnd mаintenаnce

72 GP2

Enhancing hardware security using hybrid CMOS/Spin Transfer Torque (STT) based Look Up Table / technology

By: Darya Almasi

Embedded Electrical and Computer Systems

Faculty Advisor: Dr. Hamid Mahmoodi

Hardware security is becoming an increasing threat to fabless semiconductor industries due to split between fabless design businesses and IC fabrication foundries that are globally distributed all over the world. / Spin Transfer Torque (STT) is a promising technology for information storage in form of magnetic rather than existing charged based memory such as SRAM, DRAM and flash. This technology due to its non-volatility, ease of programming, scalability and standard CMOS compatibility, offers a unique opportunity for enhancing the hardware security in a power and performance efficient manner. / In this research, we propose a unique application for STTRAM and that is to realize reconfigurable logic using Look-Up table (LUT) based logic implementation. Implementing part of a logic circuit in a programmable form hides the identity of the hardware from hacker who may attempt to reverse engineer a product to get access to intellectual property of design.

73 GP2

Wireless Autonomous Car

By: Ravi Teja Mamidipaka, Luis Lorenzo Bill Clark, and Siddharth Jankar

Embedded Electrical and Computer Systems

Faculty Advisor: Dr. Xiaorong Zhang

As part of our proposed project solution, our group has modified an RC toy car and convert it into our test platform for the project. The RC Toy Car will be modify to use an off shelf motor driver, which will allow us to control the direction of the movement of the car, as well as the speed. Then, the Tiva C Series LaunchPad will be installed on the RC Car Toy, and it will be used as the main brain of the RC Toy Car. The communication will then be obtained through wireless connectivity. The group proposes wireless connectivity because a GUI (Graphical User Interface) is going to be used for both, control the car manually, as well as receive and visualize data coming from the LaunchPad. For this task, our group will use Kivy, a Python based GUI library. It allows for easy implementation of TCP/IP or XML-RPC communication (communication protocols that can be used with Wi-Fi).

74 GP2

Stutter Controlling Device

By: Sarah Shamsi

Embedded Electrical and Computer Systems

Faculty Advisor: Dr. Xiaorong Zhang

Stuttering is a speech disorder, as reported by many adult stutterers often influenced by emotional reactions. However, the nature of the relation between stuttering and emotions is still unclear, and thus appropriate treatment is currently not available for addressing this factor. This project aims to apply advanced computer and engineering technologies to the stuttering research to enhance the field using multidisciplinary solutions; and use emerging wearable technologies by using different types of physiological sensors. This will give the comprehensive relation between stuttering, emotions, and physiological changes and propose new anti-stuttering assistive method for real-time identification and elimination of stuttering-related anticipatory anxiety.

75 GP2

Power/Area Efficient Circuits of Neuromorphic Computing System

By: Weijie Zhu, Kang Jun Bai, and Shi Jie Chen

Embedded Electrical and Computer Systems

Faculty Advisor: Dr. Hao Jiang

The neuromorphic system has the potential to be widely used in a high-efficiency artificial intelligence system, such as the optical character recognition (OCR). The development of the memristor technology provides a promising technological path to realize a high-efficient neuromorphic system. However, the readout circuit that is used to interface rest of the neural circuits is made of operational amplifiers, with power-hungry, slow and sensitive to noise. This project is to develop an integrate-and-fire circuit, an operational amplifier circuit, and a digital to analog circuit to improve the performance of readout circuit. These circuits are based on 0.13um technology of integrated circuit design.

76 GP2

Reliability Analysis of Spin Transfer Torque based Look up Tables under Process Variations and NBTI Aging

By: Ragh Kuttappa

Embedded Electrical and Computing Systems

Faculty Advisor: Dr. Hamid Mahmoodi

Spin Transfer Torque (STT) switching realized using a Magnetic Tunnel Junction (MTJ) device has shown great potential for low power and non-volatile storage. A prime application of MTJs is in building non-volatile Look Up Tables (LUT) used in reconfigurable logic. Such LUTs use a hybrid integration of CMOS transistors and MTJ devices. This paper discusses the reliability of STT based LUTs under transistor and MTJ variations in nano-scale. The sources of process variations include both the CMOS device related variations and the MTJ variations. A key part of the STT based LUTs is the sense amplifier needed for reading out the MTJ state. We compare the voltage and current based sensing schemes in terms of the power, performance, and reliability metrics. Based on our simulation results in a 16nm CMOS, for the same total device area, the voltage mode sensing scheme offers 75% lower failure rates under threshold voltage (Vth) variations, 4.9X higher tolerance to MTJ resistance variations, 19% less delay, and 64% lower active power compared to the current sensing scheme. Moreover, we compare the reliability of the two sensing schemes under Negative Bias Temperature Instability (NBTI) of PMOS transistors. Our results indicate that the voltage mode sensing scheme exhibits significantly higher tolerance to NBTI aging as well.

77 GP2

Hardware Implementation of Edge Detection and Feature Extraction for Real Time Applications

By: Mehrdad Mahdavi

Embedded Electronics and Computer Systems

Faculty Advisor:

The main challenge in real time edge detection in image recognition with software is, speed. While in one second we need billions of calculations to be performed, software's performance is completely relying on the hardware (typically CPU) which is running on it. / Design and dedicating of an ASIC whose ALU is solely arranged for a specific application will extensively improve this deficiency in a way that for example we will not have a pause in the stream of a high resolution video input

78 GP2

Variable Importance in Micro-Environment Based Protein Functional Analysis

By: Arthur Vigil

Computer Science

Faculty Advisor: Dr. Dragutin Petkovic, Dr. Kazunori Okada, and Mike Wong

Machine learning methods have been used with some success to classify protein function based on structure. Random Forest is one such method which uses the behavior of a "forest" of randomly seeded decision trees to make a classification. We apply a variant of the Random Forest method to the problem of protein function classification in order to identify structural features that make the best predictors for a particular functional model.

79 GP2

Privacy monitoring in Firefox OS

By: Harsha Cheruku and Sammy Patenotte

Computer Science

Faculty Advisor: Dr. Arno Puder

Popular mobile platforms such as iOS and Android are capable to collect
and access user's private data including current location, address
books, or other sensitive information. In many cases it is not
transparent to the user when an app accesses which kind of information.
In this project we have enhanced Mozilla's smartphone operating system,
called Firefox OS, to log access to privacy sensitive information. The
user is able to monitor the behavior of each app to recognize possible
break of privacy.

80 GP2

Smart Read

By: Imran Alavi

Computer Science

Faculty Advisor: Dr. Ilmi Yoon

Reading behavior in human beings has been an area of research for a long time and researchers are unable to completely determine all the parameters which affect it. This goal of this project is to develop a new reading model which attempts to study and research different dimensions of the reading behavior along with developing new tools to aid the reader.

81 GP2

Global multiple network alignment by combining pairwise network alignments

By: Juris Puchin and Jakob Dohrmann

Computer Science

Faculty Advisor: Dr. Rahul Singh

Background: A wealth of protein interaction data has become available in recent years, creating an / urgent need for powerful analysis techniques. In this context, a problem of particular interest is of / finding biologically meaningful correspondences between different protein-protein interaction / networks (PPIN). The PPIN of a species can be compared with that of other species through PPIN / alignment, providing insight into fundamental problems like species evolution and network / component function determination, as well as translational problems such as target identification / and elucidation of mechanisms of disease spread. Furthermore, multiple PPINs can be aligned / simultaneously, increasing the analytical implications of this formulation. While there are several / pairwise network alignment algorithms available, currently few are capable of multiple network / alignment (MNA). / / Results: We propose SMAL, a MNA algorithm based on the philosophy of scaffold-based alignment, / which is capable of converting results from any global pairwise alignment algorithms into a MNA in / linear time. Using the proposed method, we have built multiple network alignments based on / combining pairwise alignments from a number of publicly available pairwise network aligners. We / tested SMAL using PPINs of eight species derived from the IntAct PPI repository. To evaluate the / performance of the proposed method, we have applied a number of metrics including the number of / aligned nodes and edges, the Schlicker functional similarity scores (FunSIM) of aligned nodes, the / homologene similarity of aligned nodes, as well as node normalized versions of these metrics. / Additionally, as part of our experimental investigations, we compared the effectiveness of SMAL / while aligning four to eight input PPINs, and examined the effect of scaffold network choice on the / obtained multiple network alignments. / / Conclusions: SMAL was able to achieve comparable performance to that of the native MNA / implementation of IsoRankN and SMETANA, with significantly faster computational time. We also / show that SMAL was able to carry the characteristics of the pairwise alignment algorithms into the / MNA, as measured by the number of nodes aligned, number of edges aligned, as well as functional / and homologene similarity of aligned nodes. The speed, flexibility and ability to maintain obtained / alignment data as additional PPINs are added (persistence) make SMAL an excellent addition to the / toolbox of any researcher interested in large MNAs.

82 GP2

Predicting Effects of Modifying Species' Parameters in an Allometric Trophic Network Model

By: Justina Cotter

Computer Science

Faculty Advisor: Dr. Ilmi Yoon

The goal of the project is to use human visual reasoning to fit an allometric trophic model to empirical data. Fitting a model involves selecting model parameter values that accurately describe predator-prey interactions for a specific ecosystem. A fitted model allows ecologists to refine their understanding of predator-prey interactions and to make predictions about the effects of disruptions to ecosystems. / This two-pronged project creates a game, the Convergence Game, using the World Of Balance game engine. Players seek to optimize species' parameters to match target ecosystem data. Additionally, the project uses data mining techniques to identify predictors of the interspecies effects of modifying parameters for a single species in a model ecosystem. It is hoped that simple predictors can be found that can be used to make recommendations to players.

83 GP2

Newspoints

By: Luv Ahuja

Computer Science

Faculty Advisor: Dr. Arno Puder

One difficult and challenging task of a journalist is to record an incident in a way that people can / / understand and interpret easily what actually happened and how it happened. Usually journalists record / / information about an incident on paper and later arrange the content they have recorded with some / / modification before publishing it. It is often strenuous and time-consuming to record every single minute / / piece of information on paper and later recall it to map that information to its corresponding incident. This / / not only adds laborious work and decreases productivity but could also restrains journalists from / / recollecting important points about the incident which can in turn potentially reduce the speed and / / accuracy of their reporting. / / To make their work a bit easier and efficient, we conceived of developing an Android application that shall / / help journalists to record the information and at the same time arrange it in order along with the incident. / / The primary goal of Newspoints is to allow journalists to take advantage of recording incidents and / / information concurrently. In addition, this application will help journalist record details of an incident with / / few clicks while at the same time eliminating the need to carry usual tools such as paper, pen, cameras etc. / / Newspoints organizes and guides producers through the digital newsgathering and reporting process. / / Notes, multimedia and interviews and research are automatically tagged, geocoded and linked in real time. / / Newspoints “guides” prompt producers through their news gathering. “Guides” provide templates for / / news reporting situations based on input from experienced reporters, editors, producers and journalism / / educators. All this information is stored in an XML file in a particular format. This XML file is passed as / / input to the Newspoints application which then parses this XML file and shows all the questions to be / / asked while conducting an interview, source name and shot type. / / All the data and multimedia collected by Newspoints are stored locally on the mobile device. All the / / information collected during the Interview is organized in a MySQL database and retrieved to show all the / / assets belonging to a particular project. A user can review all of the assets which have been collected / / during the reporting process. The review screen displays clip metadata. User can delete any clip by long / / pressing on the asset. / / Newspoints gives users the option of archiving their assignment to and from remote storage such as / / Dropbox or Google Drive.

84 GP2

Severity Quantification of Pediatric Viral Respiratory Illnesses in Chest X-ray Images

By: Marzieh Golbaz and Bardhyl Ymeri

Computer Science

Faculty Advisor: Dr. Kazunori Okada

Early and accurate assessment of severity of viral respiratory illnesses (VRIs) allows early interventions to prevent morbidity and mortality in young children. This study proposes a novel imaging biomarker framework with chest x-ray images to assess the severity of VRIs in infants, developed specifically to meet the distinct challenges for pediatric population. The proposed framework integrates two novel technical contributions: a) obtrusive object removal using graph cut segmentation with asymmetry constraint, and b) severity quantification using information-theoretic heterogeneity measures. The experimental results with a dataset of 148 images and the ground truth severity scores given by a board-certified pediatric pulmonologist, demonstrates the effectiveness and clinical relevance of the presented framework.

85 GP2

Learning From Imbalanced dataset

By: Mehari Weldetsion and Jeffrey Hung

Computer Science

Faculty Advisor: Dr. Kazanori Okada

Imbalanced data is one of the major and common problems in machine learning (ML) practices. When a target event to be learned occurs only rarely, the number of examples available for the target will be much smaller than those for other events, making the number of training samples for each class largely different. When this skew between the minority class and the majority classes become very large, most conventional ML methods would fail to correctly account for the minority class, biased by the majority data. The objective of this study is to investigate currently available dataset balancing techniques and come up with best balancing solution for our Protein crystallization images dataset whereby the classification performance of the model would be improved. This study is significant since the use of an efficient ML image classifier could be very helpful in automating the tedious and error prone task of manual image inspection.

86 GP2

Ebluna -- 3rd Party Customer Service Software for Mid-Sized Ebay Sellers

By: Nathan Luis

Computer Science

Faculty Advisor: Dr. Ilmi Yoon

Millions of items are sold on the Ebay platform every day. There exists numerous 3rd party software to help sellers list items, ship items... However, there is a lack of software devoted to the Ebay process after the sale. Most customer service actions can only be done through the Ebay web portal, however, it is not a user friendly solution. In this project, I will create a 3rd party software targeted towards the customer service process, which occurs after the sale. The Ebluna software will utilize Java/JavaFX for the front end, a SQLite database, and Ebay webservices on the backend. The goal is to streamline the customer service aspect of selling on the Ebay platform.

87 GP2

DO IT YOURSELF: Burglar Alarm

By: Pratik Jaiswal

Computer Science

Faculty Advisor: Dr. Arno Puder

Home Automation helps managing appliances around the house. Low-powered micro-controllers allow customized functionality for which only expensive special-purpose products are available. In this project, we demonstrate how a DIY Burglar Alarm can be built using parts for less than $30" /

88 GP2

Flocking Behavior

By: Robert Moon

Computer Science

Faculty Advisor: Dr. Ilmi Yoon

I will show the use of simple forces in moving objects on a computer screen that will create emergent behavior, like the flocking of birds or a shoal of fish.

89 GP2

Structuring Unstructured Clinical Narratives in OpenMRS with Medical Concept Extraction

By: Ryan Eshleman

Computer Science

Faculty Advisor: Dr. Barry Levine, Dr. Hui Yang, and Dr. Anagha Kulkarni

In this project, we perform an extensive empirical evaluation of four Named Entity Recognition (NER) systems using textual clinical narratives and full journal biomedical articles. The results of this evaluation are leveraged to implement a module in the open source medical records system OpenMRS to perform NER functions. The four NER systems under evaluation are the National Library of Medicine’s MetaMap , Apache cTAKES, University of Michigan’s MGrep, and Arizona State University’s BANNER. Both MetaMap and cTAKES employ Maximum Entropy based chunking followed by a dictionary lookup, while BANNER uses a pure Conditional Random Fields implementation. MGrep is closed source, and its mechanism is not explicitly known. Furthermore, we have also extensively studied different ensemble approaches built upon the above four NER systems to exploit their collaborative strengths. Evaluations are performed using the hand annotated patient discharge summaries provided by the Informatics for Integrating Biology and the Bedside group (I2B2). We also evaluate these NER systems using the CRAFT dataset. Evaluation measurements include the standard Precision, Recall and F1 based on three span-based matching criteria (exact match, single boundary, any overlap), with or without requiring matching the semantic type of an entity. The main results include (1) BANNER significantly outperforms the other three systems on the I2B2 dataset with F1 values in the range of .73-.89, in contrast to .28 - .60 of other systems; (2) cTAKES, MetaMap and MGrep demonstrate different strengths and weaknesses at identifying different types of medical entities/concepts. An ensemble approach of these three systems in general outperforms each individual system on the least restrictive quality measurements by .03- .07 in F1; (3) Surprisingly, an ensemble approach of BANNER with any combinations of the other three approaches tends to degrade the performance by .08 - .11 in F1 when being evaluated on the I2B2 dataset; and finally, (4) our evaluation on the CRAFT dataset indicates that BANNER might not be effective at dealing with full-length journal articles due to its highly demanding training phase.

90 GP2