16484 research outputs found
Sort by
Predictive modeling of novel-material antennas through complex dielectric measurement
No full textThis work aims to explore a method for characterizing lossy, non-ideal conductors for use in antenna applications given the myriad emerging materials and fabrication techniques. A previously published paper details the use of onion-like carbon (OLC) and multi-walled carbon nanotube (MWCNT) films as radiating dipole antennas in the ubiquitous 2.4GHz frequency band (1). Though exhibiting clearly functional devices, the results were not predictable. Studies have attempted to characterize these materials for electronics applications, though at significantly lower frequencies pertaining to supercapacitor charge and discharge applications (2) (3) (4). These devices generally operate in the kilohertz regime, at frequencies orders of magnitude lower than the GHz regime relevant to the majority of modern communications devices. A material characterization technique is demonstrated and used to create a model to facilitate finite element simulation of carbon film antennas. The results of these finite element simulations from Ansoft High Frequency Structural Simulator (HFSS) are compared with the previously published OLC antenna results (1). Additional iterations of the model are then executed with varying dipole geometries and its efficacy analyzed versus measurements of antennas fabricated with these same dimensions. Results analysis shows excellent agreement in the fundamental frequency of antennas simulated with this model versus experimentally measured values.Ph.D., Electrical Engineering -- Drexel University, 201
Control of Small Magnetic Object in Artificial Human Tissue Material
No full textThe increasing availability of microbiome survey data has led to the use of complex machine learning and statistical approaches to measure taxonomic diversity and extract relationships between taxa and their host or environment. Accurately representing microbiome community structure has notable implications in medicine because recent work has demonstrated bidirectional interplay between microbiota and various organ systems. However, many approaches inadequately account for difficulties inherent to microbiome data, such as (1) insufficient sequencing depth resulting in sparse count data, (2) a large feature space relative to sample space, resulting in data prone to overfitting, and (3) library size imbalance, requiring normalization strategies that lead to compositional artifacts. Still, there exist approaches from other domains (e.g., natural language processing) that may be well-equipped at fitting microbiome data and may provide meaningful features that capture relevant aspects of the data. Two methods in particular are topic models and word embeddings, which characterize word co-occurrence as topics and capture semantic and lexical information of each word based on the word's neighbors, respectively. In this work, we show that a topic model can represent microbiome abundance data as topics, capturing ``subcommunity'' structure from co-occurrence patterns among taxa, whereas word embeddings can represent a nucleotide subsequence as a dense, numeric vector that encapsulates the nucleotide neighborhood in which the subsequence exists. Specifically, we present two approaches, both of which are applied to 16S rRNA amplicon surveys. First, we utilize a topic model approach. We show that library-size normalization is unnecessary and, by exploiting topic-to-topic correlations, the topic model can successfully capture complex signals such as dynamic time-series behavior of taxonomic subcommunities. In addition, we present themetagenomics to demonstrate that topic features are flexible for downstream analysis. We link taxonomic co-occurrence to their predicted functional content by leveraging gene function prediction algorithms and a fully Bayesian multilevel regression model. Second, we use Skip-Gram word2vec and a recent sentence embedding approach to embed nucleotide sequences. Our results show that embedding sequences results in meaningful representations that can be used for exploratory analyses or for downstream machine learning applications that require numeric data. The sequence embeddings can preserve relevant information about the sequencing data such as k-mer context, sequence taxonomy, and sample class. The insights we provide are applicable to various types of count data that extend beyond the microbiome sequencing domain. These include ecological presence/absence surveys, RNAseq gene expression studies, metagenomic or whole genome sequencing studies, proteomic or metabolic research, text-based studies, and econometrics. In addition, our approaches for exploring the sequence embedding space are applicable to any type of text-base research, including genetics and natural language processing, as well applications utilizing deep learning, where embedding layers are used to encode text for deeper layers of the network. Lastly, our simulation approaches and evaluation of normalization techniques are generalizable, such that aspects of these strategies could be applied to microbiome studies and work consisting of compositional data other than 16S rRNA amplicon surveys.M.S., Electrical Engineering -- Drexel University, 201
Markov Models for Procedural Content Generation
Get PDFProcedural content generation (PCG) is a growing area of research focused on leveraging artificial intelligence in the design and creation of content (e.g., levels, environments, stories, etc.) oftentimes for video games. However, most current PCG approaches are domain specific or require a substantial amount of domain knowledge to be used across multiple domains. We want to determine whether more general approaches to PCG are possible (i.e., approaches that can be applied across large classes of domains without customization or domain knowledge). The first key contribution of this dissertation is to show that machine learning approaches, specifically Markov models, can be used to model and generate levels across multiple domains by replacing domain knowledge with training data, while still being able to capture much of the domain information, such as structural level information and player interactions. The second key contribution of our work is a new theoretical framework to understand PCG approaches based on machine learning, and provide a unifying view of this new class of approaches, highlighting similarities, differences, and providing insights into future avenues of research. Our third main contribution is the development of extensions to these machine learning-based approaches that allow for more control over the generated content and more accurate modeling of the given domain.Ph.D., Computer Science -- Drexel University, 201
Fundamental Limits of Communication in Distributed Computation Frameworks
No full textThis dissertation develops a method for integrating information theoretic principles in distributed computing frameworks, distributed learning, and database design. In particular, it incorporates compression coding in such a way as to accelerate the computation of statistical functions of the data in distributed computing frameworks. Moreover, it studies the limits of decentralized compressors for query processing which speeds up computations by requiring less network traffic and efficient memory usage, and proposes a rate-distortion principle to gain some insight in distributed learning algorithms. The dissertation's insights into tradeoffs in practical distributed computing are garnered by an information theoretic model for decentralized function computation. In decentralized function computation, a series of two or more networked nodes, each storing or measuring local observation sequences, separately encode their observations, and communicate interactively with a sink, hence-forth the central estimation officer (CEO), with the goal of enabling it to compute some function or statistic of their combined data across each element in the sequence. Of great interest in this problem is the rate-distortion region, describing all possible vectors of rates, determining the sizes of each of the messages sent for which there exist encoders at the observation nodes and a decoder not exceeding a specified distortion. We assemble this problem formulation to model the cache formation design in distributed shared memory systems for query processing. Unlike traditional designs that focus on a particular query (e.g., search, retrieval, similarity), we aim to couple the representation extraction with compression and generalize the query to any types of function. We study the single letter description of the achievable region of trade-off between the storage size versus accuracy (general distortion metric) and storage size versus the cost of cache miss (log-loss distortion) for finite-alphabet sources. Another main source of difficulty after obtaining the rate-distortion expression is finding practical schemes and computing the region. The second part of the dissertation is focused on designing optimal encoding functions. With the aim of enabling the master node to compute the extrema function, the fundamental lower bound on the information exchange rate required over all quantization schemes, both scalar and vector, is computed for this interactive problem with a known iterative convex geometric method. Next, an optimal dynamic program achieving the minimum expected rate and expected rate delay tradeoff over all scalar quantization schemes is presented, and the benefits of enabling nodes to overhear each others' messages is assessed. Furthermore, a series of substantially reduced complexity dynamic programs are shown, both theoretically and empirically, to obtain performance close to the fundamental limits, and to scale favorably as the number of nodes grow. The final chapter of the dissertation proposes a novel method to design practical codes for the dissertation's accelerated distributed computing model by interactively training multi-task neural networks. This deep learning model is able to design the memory content to maximize cache hits even if the query is not known in advance. This use of deep neural networks to design codes for interactive distributed cache formation is inspired by a popular recent insight, known as the information bottleneck, that activations of the nodes at each layer of a neural network can be reinterpreted as a sufficient statistic compressing the information provided by the previous layer about the ultimate classification problem. Multiple experiments show that our neural cache formation machine can learn to compute a set of possible query functions from the cache contents by caching a learned meaningful representation of the data universal across these query tasks.Ph.D., Electrical Engineering -- Drexel University, 201
Parent Involvement in Early Childhood Education and its Impact on the Development of Early Language and Literacy Skills: An Exploration of One Head Start Program's Parent Involvement Model
Get PDFThe purpose of this qualitative study was to examine parent involvement practices of the Early Beginnings Head Start program and to determine the impact that parent involvement has on the early language and literacy development of preschool children. Although parent involvement practices are evident throughout Head Start programs, the specific practices and the effectiveness of early language and literacy acquisition from one preschool site to another may differ. The specific research questions examined in this study were: 1. How does the Early Beginnings Head Start program support families in understanding early childhood development in the areas of language and literacy development? 2. What family support models does the Early Beginnings Head Start program use to assist families in early childhood development, specifically in language and literacy skills? 3. How are parents and families utilizing these strategies at home? Data sources utilized for this research study include: (a) semi-structured interviews with a Head Start program administrator, preschool teachers and home visitors, and a small subset of parents; (b) surveys given to all parents and families in the part-time preschool program; (c) researcher field notes and observations, and (d) artifacts collected that represent parent-school interaction. Content analysis procedures were used to analyze the data, collect and identify patterns and trends from both the survey and interview data, and to determine to what extent (if any) the parent involvement practices at Early Beginnings Head Start contribute to the development of early language and literacy skills in preschool children. This research study revealed that the Early Beginnings program utilizes several different methods of involving and educating parents including the use of workshops, home visits, conferences, and newsletters. The data demonstrated that workshops were the most valuable method of parent involvement used by the program to educate parents regarding early language and literacy skills. The use of multilingual facilitators, hands-on learning, and discussions at the workshop proved to benefit parents greatly. The parent involvement methods used by Early Beginnings that occurred during this research study had varying degrees of impact on parents and the ways they facilitate learning in the home.Ed.D., Educational Leadership and Management -- Drexel University, 201
User Experience Evaluation of a Virtual Reality Prototype for Training Accuracy in Sports
No full textDifferent sports require accuracy when throwing a ball (i.e. basketball, baseball, American football), but we only rely in our judgement to guide us or correct us when doing these tasks. As we are humans, these judgements are based on imprecise information that disappear immediately from our minds, but technology can help us by showing interactive guides based on exact calculations. This project created a prototype using Virtual Reality and Motion Capture to achieve this goal, and it was evaluated by 11 users. This helped us determine that the tool has potential to be a valid solution since the users did not feel high discomfort and they considered the guides useful. But improvements need to be done in the way the feedback of the performance is provided, and in the view by making it in Augmented Reality.M.S., Digital Media -- Drexel University, 201
The effect of metal halides on MoS2 growth by chemical vapor deposition
No full textTwo-dimensional (2D) materials are materials in which the crystal structure is arranged only along two dimensions. The first 2D material to be fabricated was graphene, and it has been studied intensively for its unique band structure and excellent mechanical properties. Transition metal dichalcogenides (TMDs) are another type of 2D material that has a sandwich structure where a layer of transition metal atoms is situated between two layers of chalcogen atoms. MoS2, an example of TMDs, is the most studied 2D material because of the transition from indirect to direct bandgap when the layer number decreases. Chemical vapor deposition (CVD) is a widely used technique to synthesize 2D MoS2. Large-scale thin films can be obtained from this method; however, a limitation of CVD is that the reaction temperature required for growth is high (~800 °C). In this work, I synthesized MoS2 on SiO2 with the assistance of different alkali metal halides (NaCl, NaF and NaBr) at reduced temperatures through CVD, and characterized the properties of the films with optical microscopy, Raman microscopy, photoluminescence and atomic force microscopy. I show that alkali metal halides can reduce the growth temperature of MoS2 from 850 °C to 750 °C, and different alkali metal halides have different effects. The thickness of MoS2 synthesized with the assistance of NaCl at 750 °C is close to bulk, while the thickness of MoS2 synthesized with the assistance of NaBr and NaF at 750 °C is monolayer. However, the size of the films synthesized with the assistance of NaF at 750 °C is much smaller than that with NaBr, indicating that NaBr is the most suitable precursor that can be used to lower the reaction temperature while maintaining the crystalline quality at the investigated conditions. By demonstrating a means to reduce the CVD growth temperature, this work has a potential to advance synthesis of MoS2 on temperature-sensitive substrates.M.S., Materials Science and Engineering -- Drexel University, 201
Transitioning to and Sustaining an Inquiry Based Pedagogy
No full textMuch has been studied and written with regard to inquiry and why this is a strong and beneficial method of instruction given the current focus on 21st century skills, critical thinking, and engagement. In addition, connections between successful schools and administrators who actively practice instructional leadership have been well documented. As instructional leaders, school administrators are tasked with the responsibility of ensuring quality practices that promote student growth and successful attainment of district standards and objectives. Building administrators play a vital role in helping teachers transition from linear, didactic approaches to those of inquiry. However, we lack an in-depth understanding of how school administrators encourage, guide, and sustain classroom instructional strategies which utilize inquiry-based pedagogy. To better understand this problem, a phenomenological study was conducted to explore the lived experiences of administrators who shared their approaches, views, and insights as they assisted teachers transition to and sustain an inquiry-based approach. Three questions served as a guide for this study: (1) How do principals describe their lived experiences in developing and implementing inquiry-based teaching experiences? (2) How do principals perceive their roles in encouraging teacher practice of inquiry methodologies? (3) How do school principals describe their lived experiences in developing strategies that contribute to the successful implementation and sustainability of inquiry-based learning?Ed.D., Education -- Drexel University, 201
Traditional Group Work Songs: Implications for Modern Music Therapy Practices
No full textPre-industrial work songs performed in groups are an area of ethnomusicological literature that has not been sufficiently examined by music therapists. This capstone sought to critically examine ethnomusicological reports of work songs and uncover underlying themes that hold significant within the music therapy sphere. A literature review of multicultural work songs was conducted and organized by themes relevant to music therapy techniques and interventions. These themes were synthesized with literature on group music making, physiological elements of rhythmicization, synchrony and rhythmic entrainment, and different approaches to music therapy. Results pointed to an overlap between work songs and music therapy techniques, suggesting that work songs can better inform disciplines such as community music therapy and neurologic music therapy. Recommendations are made for developing music therapy interventions, working with different client populations and in different treatment settings, and improving music therapy education and multicultural competence. Limitations in the music therapy field and the scope of the capstone are discussed and directions for future research are presented.M.A., Music Therapy and Counseling -- Drexel University, 201
Mindful Art Making for Helping Professionals: An Art Therapy Method
No full textThis capstone describes the development and implementation of a method that combined mindfulness and art therapy to address the stress faced by helping professionals. The method intended to offer a therapeutic mindful art making experience as a means of self-care and self-compassion in a group open studio format. Observations and feedback from participants suggest that the process elicited relaxation, reflection, and self-efficacy and encouraged a culture of self-care and self-compassion. The experience led to insights and recommendations that may help guide research and implementation in the future.M.A., Art Therapy and Counseling -- Drexel University, 201