Engineering Commons^™

Generative Language Models For Personalized Information Understanding, Pengshan Cai

Doctoral Dissertations

A major challenge in information understanding stems from the diverse nature of the audience, where individuals possess varying preferences, experiences, educational and cultural backgrounds. Consequently, adopting a one-size-fits-all approach to provide information may prove suboptimal. While prior research has predominantly focused on delivering pre-existing content to users with potential interests, this thesis explores generative language models for personalized information understanding. By harnessing the potential of generative language models, our objective is to generate novel personalize content for individual users. As a result, users from diverse backgrounds can be provided with content that are tailored for their need and better aligns …

Singular Integration By Interpolation For Integral Equations, Ioannis Kyriakou

Doctoral Dissertations

Maxwell’s equations and the laws of Electromagnetics (EM) govern a plethora of electrical, optical phenomena with applications on wireless, cellular, communications, medical and computer hardware technologies to name a few. A major contributor to the technological progress in these areas has been due to the development of simulation and design tools that enable engineers and scientists to model, analyze and predict the EM interactions in their systems of interest. At the core of such tools is the field of Computational Electromagnetics (CEM), which studies the solution of Maxwell’s equations with the aid of computers. The advances in these applications technologies, …

Sustainable Computing - Without The Hot Air, Noman Bashir, David Irwin, Prashant Shenoy, Abel Souza

Publications

The demand for computing is continuing to grow exponentially. This growth will translate to exponential growth in computing's energy consumption unless improvements in its energy-efficiency can outpace increases in its demand. Yet, after decades of research, further improving energy-efficiency is becoming increasingly challenging, as it is already highly optimized. As a result, at some point, increases in computing demand are likely to outpace increases in its energy-efficiency, potentially by a wide margin. Such exponential growth, if left unchecked, will position computing as a substantial contributor to global carbon emissions. While prominent technology companies have recognized the problem and sought to …

Benchmarking Small-Dataset Structure-Activity-Relationship Models For Prediction Of Wnt Signaling Inhibition, Mahtab Kokabi

Masters Theses

Quantitative structure-activity relationship (QSAR) models based on machine learning algorithms are powerful tools to expedite drug discovery processes and therapeutics development. Given the cost in acquiring large-sized training datasets, it is useful to examine if QSAR analysis can reasonably predict drug activity with only a small-sized dataset (size < 100) and benchmark these small-dataset QSAR models in application-specific studies. To this end, here we present a systematic benchmarking study on small-dataset QSAR models built for prediction of effective Wnt signaling inhibitors, which are essential to therapeutics development in prevalent human diseases (e.g., cancer). Specifically, we examined a total of 72 two-dimensional (2D) QSAR models based on 4 best-performing algorithms, 6 commonly used molecular fingerprints, and 3 typical fingerprint lengths. We trained these models using a training dataset (56 compounds), benchmarked their performance on 4 figures-of-merit (FOMs), and examined their prediction accuracy using an external validation dataset (14 compounds). Our data show that the model performance is maximized when: 1) molecular fingerprints are selected to provide sufficient, unique, and not overly detailed representations of the chemical structures of drug compounds; 2) algorithms are selected to reduce the number of false predictions due to class imbalance in the dataset; and 3) models are selected to reach balanced performance on all 4 FOMs. These results may provide general guidelines in developing high-performance small-dataset QSAR models for drug activity prediction.

Natural Language Processing For Lexical Corpus Analysis, Abram Kaufman Handler

Doctoral Dissertations

People have been analyzing documents by reading keywords in context for centuries. Traditional approaches like paper concordances or digital keyword-in-context viewers display all occurrences of a single word from a corpus vocabulary amid immediately surrounding tokens or characters, to show readers how individual lexical items are used in bodies of text. We propose that these common tools are one particular application of a more general approach to analyzing documents, which we define as lexical corpus analysis. We then propose new natural language processing techniques for lexically-focused corpus investigation, and demonstrate how such methods can be used to create new user-facing …

Cost-Efficient Resource Provisioning For Cloud-Enabled Schedulers, Lurdh Pradeep Reddy Ambati

Doctoral Dissertations

Since the last decade, public cloud platforms are rapidly becoming de-facto computing platform for our society. To support the wide range of users and their diverse applications, public cloud platforms started to offer the same VMs under many purchasing options that differ across their cost, performance, availability, and time commitments. Popular purchasing options include on-demand, reserved, and transient VM types. Reserved VMs require long time commitments, whereas users can acquire and release the on-demand (and transient) VMs at any time. While transient VMs cost significantly less than on-demand VMs, platforms may revoke them at any time. In general, the stronger …

Thermoelectric Transport In Disordered Organic And Inorganic Semiconductors, Meenakshi Upadhyaya

Doctoral Dissertations

The need for alternative energy sources has led to extensive research on optimizing the conversion efficiency of thermoelectric (TE) materials. TE efficiency is governed by figure-of-merit (ZT) and it has been an enormously challenging task to increase ZT > 1 despite decades of research due to the interdependence of material properties. Most doped inorganic semiconductors have a high electrical conductivity and moderate Seebeck coefficient, but ZT is still limited by their high lattice thermal conductivity. One approach to address this problem is to decrease thermal conductivity by means of alloying and nanostructuring, another is to consider materials with an inherently low …

Mtemp: An Ambient Temperature Estimation Method Using Acoustic Signal On Mobile Devices, Hao Guo

Masters Theses

Ambient temperature sensing plays an important role in a number of applications in agriculture, industry, daily health care. In this thesis project, we propose a new acoustic-based ambient temperature sensing method called Mtemp. Mtemp empowers acoustic-enabled IoT devices, smartphones to perform ambient air temperature sensing without additional hardware. Basically, Mtemp utilizes on-board speaker and microphone to calculate the propagation speed of acoustic signal by measuring the phrase of the target signal, thereby estimate the ambient temperature according to a roughly linear relationship between temperature and sound speed. Mtemp is portable and economical, making it competitive compared with traditional thermometers for …

Real-Time Tddft-Based Filtered Spectroscopy, Ivan Williams

Masters Theses

We demonstrate the ability to simulate targeted excitation of atomistic systems using our real-time TDDFT-based simulation framework NESSIE. Traditional ap- proaches for extracting spectra through real-time TDDFT involves excitation of all frequencies via impulse requiring long simulation times and yielding poor resolution. By exciting the system with a modulated sinc pulse between the frequencies of inter- est we are able to obtain a spectral response with far more precision in a significantly shorter time frame than competing implementations.

The Effect Of Oxygen On Properties Of Zirconium Metal, Jie Zhao

Doctoral Dissertations

The influence of oxygen on the thermophysical properties of zirconium has been investigated using MSL-EML (Material Science Laboratory Electromagnetic Levitator) on ISS (International Space Station) in collaboration with NASA (National Aeronautics and Space Administration), ESA (European Space Agency), and DLR (German Aerospace Center). Zirconium samples with different oxygen concentrations was subjected to multiple melt cycles during which the thermophysical properties, such as density, viscosity and surface tension, have been measured at various undercooled and superheated temperatures. Also, there are melt cycles for verifying the solidification mechanism. Similar samples were found to show anomalous nucleation of the solid for certain ranges …

Computational Modeling Of Laminated Veneer Bamboo (Lvb) Dowel Joints, Niloufar Khoshbakht

Doctoral Dissertations

Laminated veneer bamboo (LVB) is a sustainable building material that has been gaining interest in the construction industry of late. As a relatively new product, little is known about its connection performance, specifically, its failure behavior in dowel type joints and possible similarities it may have to engineered wood products in terms of failure mechanisms. Research is needed to aid in the understanding of LVB dowel connection failure behavior and to quantify the failure mechanism and key factors associated with LVB dowel connection strength. Modeling, as conducted in this research, is a valuable tool to help devise safe standards and …

A Multiline Anchor Concept For Floating Offshore Wind Turbines, Casey Fontana

Doctoral Dissertations

Floating offshore wind turbines (FOWTs) hold great potential for the renewable energy industry, but capital costs remain high. In efforts to increase FOWT substructure efficiency and reduce costs, this thesis investigates a novel multiline anchor concept in which FOWTs share anchors instead of being moored separately. The goal of this thesis is to evaluate the force dynamics, design, and potential cost reduction of the system. Anchor forces are simulated using the NREL 5 MW reference turbine and OC4-DeepCwind semisubmersible platform, and multiline anchor force is computed as the vector sum of the contributing mooring line tensions. The use of a …

Reward Allocation For Maximizing Energy Savings In A Transportation System, Adewale O. Oduwole

Masters Theses

Transportation has a major impact on our society and environment, contributing 70% of U.S petroleum use, 28% of U.S. greenhouse gas (GHG) emissions, over 34,000 fatalities and 2.2 million injuries in 2013. Punitive approaches to used to tackle environmental issues in the transportation sector, such as congestion pricing have been well documented, although the use of incentives or rewards lags behind in comparison. In addition to the use of more fuel-efficient, alternate energy vehicles and various other energy reduction strategies; energy consumption can be lowered through incentivizing alternative modes of transportation. This paper focused on modifying travelers’ behavior by providing …

Supporting Engineering Design Of Additively Manufactured Medical Devices With Knowledge Management Through Ontologies, Thomas Hagedorn

Doctoral Dissertations

Medical environments pose a substantial challenge for engineering designers. They combine significant knowledge demands with large investment for new product development and severe consequences in the case of design failure. Engineering designers must contend with an often-chaotic environment to which they have limited access and familiarity, a user base that is difficult to engage and highly diverse in many attributes, and a market structure that often pits stakeholders against one another. As medical care in general moves towards personalized models and surgical tools towards less invasive options emerging manufacturing technologies in additive manufacturing offer significant potential for the design of …

Magneto-Electric Approximate Computational Framework For Bayesian Inference, Sourabh Kulkarni

Masters Theses

Probabilistic graphical models like Bayesian Networks (BNs) are powerful artificial-intelligence formalisms, with similarities to cognition and higher order reasoning in the human brain. These models have been, to great success, applied to several challenging real-world applications. Use of these formalisms to a greater set of applications is impeded by the limitations of the currently used software-based implementations. New emerging-technology based circuit paradigms which leverage physical equivalence, i.e., operating directly on probabilities vs. introducing layers of abstraction, promise orders of magnitude increase in performance and efficiency of BN implementations, enabling networks with millions of random variables. While majority of applications with …

Skynet: Memristor-Based 3d Ic For Artificial Neural Networks, Sachin Bhat

Masters Theses

Hardware implementations of artificial neural networks (ANNs) have become feasible due to the advent of persistent 2-terminal devices such as memristor, phase change memory, MTJs, etc. Hybrid memristor crossbar/CMOS systems have been studied extensively and demonstrated experimentally. In these circuits, memristors located at each cross point in a crossbar are, however, stacked on top of CMOS circuits using back end of line processing (BOEL), limiting scaling. Each neuron’s functionality is spread across layers of CMOS and memristor crossbar and thus cannot support the required connectivity to implement large-scale multi-layered ANNs.

This work proposes a new fine-grained 3D integrated circuit technology …

Explorations Into Machine Learning Techniques For Precipitation Nowcasting, Aditya Nagarajan

Masters Theses

Recent advances in cloud-based big-data technologies now makes data driven solutions feasible for increasing numbers of scientific computing applications. One such data driven solution approach is machine learning where patterns in large data sets are brought to the surface by finding complex mathematical relationships within the data. Nowcasting or short-term prediction of rainfall in a given region is an important problem in meteorology. In this thesis we explore the nowcasting problem through a data driven approach by formulating it as a machine learning problem.

State-of-the-art nowcasting systems today are based on numerical models which describe the physical processes leading to …

The Impact Of Quantum Size Effects On Thermoelectric Performance In Semiconductor Nanostructures, Adithya Kommini

Masters Theses

An increasing need for effective thermal sensors, together with dwindling energy resources, have created renewed interests in thermoelectric (TE), or solid-state, energy conversion and refrigeration using semiconductor-based nanostructures. Effective control of electron and phonon transport due to confinement, interface, and quantum effects has made nanostructures a good way to achieve more efficient thermoelectric energy conversion. This thesis studies the two well-known approaches: confinement and energy filtering, and implements improvements to achieve higher thermoelectric performance. The effect of confinement is evaluated using a 2D material with a gate and utilizing the features in the density of states. In addition to that, …

Eulerian Cfd Modeling Of Multiphase Internal Injector Flow And External Sprays, Eli T. Baldwin

Doctoral Dissertations

The improvement of combustion systems which use sprays to atomize liquid fuel requires an understanding of that atomization process. Although the secondary break up mechanisms for the far-field of an atomizing spray have been thoroughly studied and well understood for some time, understanding the internal nozzle flow and primary atomization on which the far-field spray depends has proven to be more of a challenge. Flow through fuel injector nozzles can be highly complex and heavily influenced by factors such as turbulence, needle motion, nozzle imperfections, nozzle asymmetry, and phase change. All of this occurs within metallic injectors, making experimental characterization …

A Combined Quadtree/Delaunay Method For 2d Mesh Generation, Simon Tang

Masters Theses 1911 - February 2014

Unstructured simplicial mesh is an integral and critical part of many computational electromagnetics methods (CEM) such as the finite element method (FEM) and the boundary element method (BEM). Mesh quality and robustness have direct impact on the success of these CEM methods.

A combined quadtree/Delunay 2D mesh generator, based on the early work of Schroeder (1991, PhD), is presented. The method produces a triangulation that approximates the original geometric model but is also topologically consistent. The key advantages of the method are: (a) its robustness, (b) ability to create a-priori graded meshes, and (c) its guaranteed mesh quality.

The method …

High Speed Flow Simulation In Fuel Injector Nozzles, Sukanta Rakshit

Masters Theses 1911 - February 2014

Atomization of fuel is essential in controlling combustion inside a direct injection engine. Controlling combustion helps in reducing emissions and boosting efficiency. Cavitation is one of the factors that significantly affect the nature of spray in a combustion chamber. Typical fuel injector nozzles are small and operate at a very high pressure, which limit the study of internal nozzle behavior. The time and length scales further limit the experimental study of a fuel injector nozzle. Simulating cavitation in a fuel injector will help in understanding the phenomenon and will assist in further development.

The construction of any simulation of cavitating …

A Numerical Study On Hydraulic Characteristics In The Ice Harbor-Type Fishway, Seojun Kim, Kwonkyu Yu, Byungman Yoon, Yoonsung Lim

Journal Articles

Recently various types of fishways have been developed and constructed in order to preserve diversity of fish species and restore the riverine ecological system. Some of the fishways in Korea, however, seem to be designed and installed without careful study on their functions and swimming characteristics of target fishes. The present study aims to elucidate proper hydraulic characteristics of the Ice Harbor-type fishway. In the present study, we analyzed the hydraulic characteristics depending on changes in the spans of walls in the Ice Harbor-type fishways, by using Flow-3D, a three dimensional numerical simulation program. Two physical experiments were performed to …

News Media Environment, Selective Perception, And The Survival Of Preference Diversity Within Communication Networks, Frank C.S. Liu, Paul E. Johnson

JITP 2011: The Future of Computational Social Science

There is a natural tension between the effects on public opinion of social networks and the news media. It is widely believed that social networks tend to harmonize opinions within them, but the presence of media may accentuate diversity by inserting discordant messages. On the other hand, in a totalitarian state where the government controls the media, social networks may mitigate the homogenizing pressure of a regime’s propaganda. The tendency of opinion to follow the “official line” may be mitigated because opponents of the government interact on a personal level and bolster one another’s views. This paper employs agent-based modeling—an …

Gpu Based Lithography Simulation And Opc, Lokesh Subramany

Masters Theses 1911 - February 2014

Optical Proximity Correction (OPC) is a part of a family of techniques called Resolution Enhancement Techniques (RET). These techniques are employed to increase the resolution of a lithography system and improve the quality of the printed pattern. The fidelity of the pattern is degraded due to the disparity between the wavelength of light used in optical lithography, and the required size of printed features. In order to improve the aerial image, the mask is modified. This process is called OPC, OPC is an iterative process where a mask shape is modified to decrease the disparity between the required and printed …

Infoextractor – A Tool For Social Media Data Mining, Chirag Shah, Charles File

JITP 2011: The Future of Computational Social Science

We present InfoExtractor, a web-based tool for collecting data and metadata from focused social media content. InfoExtractor then provides this data in various structured and unstructured formats for easy manipulation and analysis. The tool allows social science researchers to easily collect data for quantitative analysis, and is designed to deliver data from popular and influential social media sites in a useful and easy to access way. InfoExtractor was designed to replace traditional means of content aggregation, such as page scraping and brute- force copying.

Researching Real-World Web Use With Roxy, A Research Proxy, Ericka Menchen-Trevino, Chris Karr

JITP 2011: The Future of Computational Social Science

Outside of a lab environment, it has been difficult for researchers to collect both behavioral and self-reported Web-use data from the same participants. To address this challenge we created Roxy, open source software that collects real-world Web-use data with participants’ informed consent. Roxy gathers Web log data as well as the text and HTML code of each page visited by participants. We describe Roxy’s data gathering capabilities and search functions and then illustrate how we used the software in a multi-method study. The use case examines selective exposure to political communication during the November 2010 U.S. general election campaign.

Tradeoffs In Accuracy And Efficiency In Supervised Learning Methods, Loren Collingwood, John Wilkerson

JITP 2011: The Future of Computational Social Science

Text is becoming a central source of data for social science research. With advances in digitization and open records practices, the central challenge has in large part shifted away from availability to usability. Automated text classification methodologies are becoming increasingly important within political science because they hold the promise of substantially reducing the costs of converting text to data for a variety of tasks. In this paper, we consider a number of questions of interest to prospective users of supervised learning methods, which are appropriate to classification tasks where known categories are applied. For the right task, supervised learning methods …

Facilitating Encounters With Political Difference: Engaging Voters With The Living Voters Guide, Deen G. Freelon, Travis Kriplean, John Morgan, W. Lance Bennett, Alan Borning

JITP 2011: The Future of Computational Social Science

Unlike 20th-century mass media, the Internet requires self-selection of content by its very nature. This has raised the normative concern that users may opt to encounter only political information and perspectives that accord with their preexisting views. This study examines the different ways that voters appropriated a new, purpose-built online engagement platform to engage with a wide variety of political opinions and arguments. In a deployment aimed at helping Washington state citizens make their 2010 election decisions, we find that users take significant advantage of three key opportunities to engage with political diversity: reading, acknowledging, and writing arguments on both …

Politics 2.0 With Facebook – Collecting And Analyzing Public Comments On Facebook For Studying Political Discourses, Chirag Shah, Tayebeh Yazdani Nia

JITP 2011: The Future of Computational Social Science

Analyzing publicly available content on various social media sites such as YouTube and Twitter, as well as social network sites such as Facebook, has become an increasingly popular method for studying socio-political issues. Such public-contributed content, primarily available as comments, let people express their opinions and sentiments on a given topic, news-story, or post, while allowing social and political scientists to extend their analysis of a political discourse to social sphere. We recognize the importance of Facebook in such analysis and present several approaches and observations of collecting and analyzing public comments from it. In particular, we demonstrate what it …

An Automated Snowball Census Of The Political Web, Abe Gong

JITP 2011: The Future of Computational Social Science

This paper solves a persistent methodological problem for social scientists studying the political web: representative sampling. Virtu- ally all existing studies of the political web are based on incomplete samples, and therefore lack generalizability. In this paper, I combine methods from computer science and sampling theory to conduct an automated snowball census of the political web and constructs an all- but-complete index of English political websites. I check the robust- ness of this index, use it to generate descriptive statistics for the entire political web, and demonstrate that studies based on ad hoc sampling strategies are likely to be biased …