The ability to form an understanding of an environment is essential for a mobile robot, enabling it to plan and execute actions efficiently. Such understanding should consist not only of geometry of space, but include notion of high-level semantic spatial concepts, such as objects, landmarks, or functional room categories. The goal of this project is to enable mobile robots to learn a unified deep generative model of the robot's environment that spans multiple levels of abstraction. Training deep models for robotics applications is often challenging. It requires large amounts of annotated data which is tedious and costly to collect in real environments. In this work, we investigate the benefits of utilizing knowledge generated in simulated environments, in which robots can be deployed significantly faster and in parallel, for improving the quality of a learned spatial model. Specifically, we generated random artificial simulated worlds and perform initial pre-training of the model there. To this end, we made following contributions. First, we improved the robustness of the navigation system of our robot DUB-E, which allowed us to deploy the robot in cluttered, human-populated environments. Second, we implemented an algorithm generating random simulated worlds for automatic data collection. Third, we designed a low-level representation of geometry of space around the robot, which serves as input to the deep spatial model. Finally, we collected and processed a large dataset of sensory information captured in simulation and the real world. We demonstrated the impact of the additional data generated in simulation on the performance of the deep spatial model.

Industries have been slow to digitize their equipment and processes. Until now, doing so has not been as accessible due to factors such as price along with the size and efficiency of the equipment with which to do so. The Internet of Things (IoT) industry is projected to grow to over 30.7 billion devices by 2020. Such a huge number of devices have the potential to become a threat to Internet security. This was seen in the attack that occurred on Oct. 21, 2016, when hackers used IoT devices to carry out large-scale DNS attacks. This example illustrates the importance of gathering and analyzing data from these IoT devices. To better understand the problem, we are implementing the use of IoT devices to gather data and populate it into a data warehouse for further analysis. We used wireless antennas to gather network data, then populated data into a data warehouse. Finally, we extracted data from this warehouse to answer questions such as, "how crowded is the wireless spectrum" and "what is the distributribution of traffic on a network"? The sensor used in this project was a Raspberry Pi paired with a USB wireless antenna. The data warehouse used an unstructured, NoSQL database. The analysis tools used were Excel and Python. The Python programming language served as the link between gathering, population and analysis of data.

Understanding the relationship between the parts of the brain that initiate behavior and those that execute movements is a fundamental goal for basic neuroscience and important for effective clinical treatment of many behavioral disorders. Electrically evoked potentials (EEPs) are a measure of the strength of neural connections between cortical sites and have been used to document connectivity both in animals and in humans. Past research has shown that there are spontaneous state-dependent changes in basal-cortical EEPs and cortico-cortical EEPs in non-human primates; the goal of this project is to examine state-dependent changes in connectivity between the prefrontal and motor cortices. We hypothesize that prefrontal cortex-motor cortex EEPs will show robust state-dependent changes, as the prefrontal cortex is heavily involved in motor planning. For this work, we collect neural data and stimulate the brain to produce EEPs using the Neurochip-3, a portable, head-mounted neurophysiology system, while the monkey is freely behaving in its home cage. Behavioral state is defined using signals from a head-mounted accelerometer and power spectral density analysis of the neural signals directly prior to stimulation events. We hypothesize that EEPs in the motor cortex evoked from the prefrontal cortex will have higher amplitude during waking movement than during rest or sleep.

Spirometry is among the most common tests for characterizing pulmonary function and plays a critical role in detecting and improving outcomes related to chronic lung disease. During a spriometry maneuver, a patient rapidly exhales into a machine which records the total volume and flow of air over time. However, patient error in performing the spirometry maneuver, such as by coughing or taking multiple breaths, can lead to clinically misleading results. As a result, spirometry must take place under the supervision of a trained specialist who can detect and correct patient errors. To reduce the need of specialists to coach patients during spirometry, we demonstrate the ability to automatically detect four common patient errors. Applying machine learning techniques on features derived from spirometry data, we were able to automatically classify errors from spirometry data with an F-score between 0.784 and 0.907. Our work provides the first step toward reducing the need for trained individuals to administer spirometry tests by demonstrating the ability to automatically detect specific errors and provide appropriate patient feedback. This will increase the availability of spirometry, especially in low resource and telemedicine contexts.

Financial inclusion - the expansion of access to formal bank accounts, particularly to the poor - has recently moved up the ranks as a priority for development. After the unexpected success of M-PESA - a mobile phone based money transfer service started in Kenya - many quickly joined the camp of mobile money advocates believing that mobile money would be the solution to this newly minted UN sustainable development goal. However, with such nuanced characteristics in each country and even more distinctions within each financial system and economy, mobile money as a solution may not be applicable in every country aiming for financial inclusion. Therefore, this research aims to primarily understand the financial practices and needs of Ghanaians. To do so, I conduct surveys in five different cities ranging from small towns to large cities to evaluate an individuals current financial practices, existing and projected needs for financial services, and the role that mobile money may play in fulfilling either of those goals. 

The Possession Sound is located in the north Whidbey Basin of the Puget Sound between Whidbey Island and Snohomish County. Due to its proximity to multiple cities and the Snohomish River, both natural and anthropogenic influences can affect the distribution of heavy metals in this area. While normal heavy metal concentrations provide benefits to the ecosystem, anthropogenic sources can result in elevated concentrations, harming the environment. Students at the Ocean Research College Academy (ORCA), a dual enrollment program through Everett Community College, have studied this issue since 2004. A ponar-type sediment grab was deployed from ORCA’s research vessel to collect sediment samples, which were analyzed by the Everett Environmental Laboratory using an inductively coupled plasma mass spectrometer (ICP-MS). This study focused on zinc and lead concentrations during 2014-2016 at three sites in the Possession Sound: Dolphin 1, Mt. Baker Terminal (MBT), and Howarth. MBT and Howarth are in close proximity to the coast, while Dolphin 1 is in the middle of Possession Sound. These sites were chosen to investigate whether coastal sites would exhibit increased zinc and lead concentrations due to anthropogenic influences. Results show that mean lead concentrations were elevated in areas closer to the coast, MBT (3.13 mg/kg) and Howarth (3.15 mg/kg), in comparison to Dolphin 1 (2.277 mg/kg). Zinc concentrations didn’t follow the same trend; Dolphin 1 had a higher average zinc concentration (27.4 mg/kg) than MBT (22.2 mg/kg) and a similar concentration to Howarth (28.5 mg/kg). While anthropogenic influences appear to dominate the distribution of lead in the Possession Sound, zinc distribution may be controlled by multiple sources. High zinc concentrations at Howarth suggest that Snohomish River outflow may be one of these sources. Future sampling at monthly intervals is necessary to determine if there is a temporal influence and explain outliers in the data.

Adverse Events (AEs) are detrimental medical incidents in a research participant that may or may not be related to the investigational agent. During clinical trials, physicians are required to document attribution of AEs with limited guidance regarding how to assign attribution. The objective of this study was to determine the accuracy of physician based attributions for AEs occurring in cystic fibrosis (CF) patients participating in blinded randomized controlled trials (RCTs). This retrospective study pooled clinical trial data from 4 CF RCTs; all participants who experienced ≥1 AE were included. A repeated measures logistic regression model using generalized estimating equations and adjusted for baseline demographic variables (gender, age, CF genotype), comorbidities, and lung function (forced expiratory volume in one second; FEV1). The primary predictor of interest was receipt of study drug versus placebo. The primary outcome of interest was physicians’ determination of the AE as related to study drug. Study patients were young with mild lung disease. Receipt of study drug was not significantly associated with the likelihood a physician would deem an AE to be related to study drug. However, we found significant associations between age, gender, and FEV1 and our outcome of interest– presence of an AE felt by a physician to be related to study drug. These findings suggest a biased assessment of physician attribution. Physician assessment of relatedness in RCT’s is a key requirement per the Food and Drug Administration (FDA). We found no significant association between AEs coded as related to study drug and actual study drug receipt. Interestingly, we found other associations implying an intrinsic bias in physician assessment based on age, gender and disease severity. Further research may clarify whether changes are needed to current FDA guidelines.

Today's problems in fields like biomedical research require complex analysis of vast amounts of data in a short window of time, and the state of the art in computer science is not up to the challenge. The status-quo approach to such computationally demanding problems has been to design a custom algorithm, but this process is difficult, time-consuming, error-prone, expensive, and rarely yields the best solution. To overcome these challenges, computer science is undergoing a paradigm shift, from solving problems with individual algorithms to solving entire domains of problems with automatic generation, or synthesis, of algorithms. In particular, many problems boil down to scheduling or, in other words, planning when to compute what. For instance, a data center must plan how to divide a given complicated task across multiple machines. When this plan or schedule is created ahead of time, the process is known as static scheduling, and this schedule constitutes an algorithm specialized to the problem. Until now, static scheduling for every new problem domain has required either a complex, custom algorithm or a complex, custom model for a logic solver. Even in simple cases, these existing approaches are highly susceptible to human misinterpretation of requirements and necessitate complex, error-prone engineering to scale to real-world usage. The situation is worse still if the schedule needs to be optimized for efficiency. My research with Professor Bodik automates the creation of scalable tools for static scheduling from simple code to interpret the meaning of a candidate schedule. This work enables application of static scheduling to sophisticated problem domains while maintaining high confidence in schedule correctness. I applied this framework to create a static scheduling tool for hierarchical computation, supporting advanced and notoriously difficult techniques like parallel and incremental computation. With this tool, I created high-performance algorithms for data visualization and web browsers.

Additive manufacturing - the process behind 3D printing - has been growing in widespread use since the introduction of the first commercially-available desktop printer and later improvements in cost, quality, and accessibility. Expanding horizontally and vertically in the global market, the 3D printing industry has been one that not only appeals to hobbyists, but has demonstrated potential as a solution to a variety of challenges faced around the world today, including those in biotechnology, aerospace, and industrial engineering. Seeking to take advantage of the increasing demands in 3D printing technologies, a number of companies and open-source organizations are offering new types of 3D printers and associated software. However, the rapid growth and diversity of the products being released have resulted in a loss of universal standards. Often indented to be accessible and reliable regardless of a user’s technical background, 3D printing technologies have been known to require consumers to learn the more technical aspects of their printer, G-Code, slicing software, materials, etc. in order to configure their overall setup and reduce the prevalence of errors. Our project aims at using advanced parsing strategies, program optimization, and statistical analysis to provide a universal, easy-to-use tool that both analyzes and optimizes G-Code files prior to printing. By automating the common fixes users are required to implement in their G-Code files and identifying common factors in 3D prints (material warping, inadequate temperatures, speed of extruder heads, etc.), this tool is intended to be usable for any consumer, regardless of engineering or design experience. Ultimately, our goal is to help provide a universal solution to the lack of consistency between products, and help leverage the promising field of 3D printing to reach its potential as both a home essential and reliable solution in fields such as biotechnology and industrial engineering.

We propose to perform an original composition for an unusual quartet of instruments that blend together very different performance traditions and practices. The piece is the result of a collaboration between musicians that normally do not work together because of a lack of common chamber music repertoire. Our composition explores the timbral and compositional possibilities of strings and both traditional and non-traditional percussion instruments. It consists of two contrasting ideas that are juxtaposed: a slow, mysterious, and lyrical idea featuring various metallic percussion and high string harmonics, and a fast mixed meter fugue. These ideas conflict and transform each other, eventually coalescing asymmetrically into a dramatic conclusion. Writing and rehearsing this piece allowed us to merge together the largely classically trained basis of string instrument performance practice with contemporary percussion performance practice, each helping inform the other and leading to unique musical ideas and results. Hopefully this piece and others like it will allow for more future collaboration between musicians from these different backgrounds and performance practices.

Duplex Sequencing (DS) is a Next-Generation Sequencing technology employing double strand molecular tagging to reduce technical errors.The high accuracy of DS enables the resolution of low frequency mutations in nuclear and mitochondrial DNA (mtDNA). We are interested in using DS to analyze mtDNA mutations in ulcerative colitis (UC), an inflammatory bowel disease that predisposes to colorectal cancer. Tumor progression in UC is related to mitochondrial dysfunction. By sequencing mtDNA mutations in colonic biopsies from UC patients, we aim to elucidate the role of these mutations in tumorigenesis. A critical limitation, however, is the high number of low frequency mtDNA mutations present in these biopsies and the lack of appropriate computational methods to process large amounts of information generated by DS. The goal of this study was to develop computational methods to enable the extraction of mtDNA mutation information from DS data. DS identifies the reads corresponding to each strand of DNA by grouping them by their unique tags, making single-strand consensus sequences (SSCSs), which are then paired to form duplex consensus sequences (DCS). SSCS reads remove technical errors, but contain errors caused by DNA damage. The first computational method that we developed compared sequencing information from SSCS and DCS reads to assess the level and type of DNA damage in each sample. Other computational methods consisted of a program to classify mutations according to their sequence context, and a script that calculates mutation types and frequencies inside and outside the D-loop region as well as in all mtDNA genes and allows comparison of these parameters in large batches of samples. These tools will enable the processing of mtDNA DS information in a fast and high-throughput way, leading to assessment of the range of different types of mtDNA mutations in the progression of ulcerative colitis and colorectal cancer.

Video games are seldom used for environmental science education, despite the variety of compelling lessons that they could be used to teach. We designed, built and tested a video game about the effects of climate change on a charismatic local animal, the pika. Native to North America and Asia, pikas are sensitive to changes in climate because they are covered with thick fur and get heat stroke even under modest temperatures (77 °F). Pikas have to gather food throughout the summer, enough for them to make it through the long winter in their burrows. When temperatures warm, pikas do not have as much time to forage for food during their day. We decided to take these ideas and turn it into the major mechanics of a game that we developed. The resulting project, Life of Pika, is a runner game in which players need to collect flowers to survive while managing their temperature to avoid overheating. We have taken inspiration from other runner games such as Frogger, Crossy Road, and Sonic the Hedgehog, but this game is unique in that player vulnerability centers around their temperature bar, rather than around avoiding obstacles. The game is divided into seven levels to represent the pika’s seven year lifespan. As one advances from level to level we make the game progressively harder by increasing the rate at which the player’s temperature increases to simulate increasing global temperatures. Developed with industry-standard software for implementation in classroom and museum settings, we aim to promote empathy in the player about the pikas’ struggle against climate change. We hope that players will become more thoughtful about their impact on the world and its inhabitants.

Viziometrics is an image search engine and classifier created by researchers at the University of Washington eScience Institute. Its purpose is to further the communication of scientific results through increased access to visual information as well as study the relationship the use of visual information and between scientific impact. Currently the search and classification are performed on 8 million images from PubMed Central, a major research database for scholarly works in the life sciences and biomedical fields. As a part of continuous efforts to improve the Viziometrics platform, we are working on a feature that automatically identifies a ‘central figure’ in a scientific publication. We have defined the central figure as either the visualization that encapsulates key aspects of a scientific publication, or that which best represents the research findings. In order to improve the underlying central figure search algorithm we are conducting a study that answers the question: which image, if any, do authors themselves identify as a “central figure” within a given publication? The research presented here is based on a survey of authors, whose email contacts can be found in the PubMed database. Based on the survey data, we are looking for common features characterizing papers for which authors were able to find ‘central figures’, as well as those for which authors could not make a decision. We are using this data to increase the accuracy of the Viziometrics algorithm. As more data sets are added to the Viziometrics platform, we hope to encourage scientists to find new ways to collaborate and communicate across disciplines.

The axons in the nervous system are collectively referred to as white matter. White matter forms intricate fiber connections among the brain’s grey matter (neurons). These connections change with development, learning, and disease, and therefore are of interest to several medical researchers. Tract-Based Spatial Statistics (TBSS) is an analysis technique which is used to analyze white matter integrity. Currently, this is the best existing method to align and analyze white matter. However, we are creating and evaluating an improved method to more accurately measure changes to integrity. Diffusion Tensor Imaging (DTI) Voxel Based Morphometry (DTIVBM) is a technique that we are developing which allows us to compare white matter statistics by using nonlinear registration to align the DTI image to a structural image and then a standard template. We are comparing several variations of DTIVBM with the established method of TBSS. We are using four different datasets, three of which contain adult brains ranging from ages 18-35, and one containing preteen/teen brains ranging from ages 11-15. One of the datasets also contains a mixture of normal brains and brains affected by neurodegenerative disease. Each subject has two separate brain images taken within a month’s gap. By running both the DTIVBM scripts and TBSS scripts on these brain images, we will be able to determine which method is more reliable, as measured using an Intra-Class Correlation (ICC) statistic. We are currently in the process of gathering the results, but we expect to see that DTIVBM will have higher reliability on most datasets than TBSS over a wider range of white matter. If this is true, we would have demonstrated a potentially better method to analyze white matter changes through aging and development, which will require fewer people and allow us to see changes in more of the brain.

This research examines the politicized conversation around officer-involved shootings. Drawing on a dataset of shooting-related tweets containing #blacklivesmatter, #bluelivesmatter or #alllivesmatter, we construct a network graph of Twitter users who participated in this conversation, utilizing a measure of shared audience to connect and cluster accounts into communities. We then employed mixed-methods, including visual and qualitative analysis, to explore the communities, influencers, and information flows of this discourse. Our findings identify and describe several distinct clusters of accounts—both left- and right-leaning clusters, including a cluster of alt-right media—and show how users within the latter cluster were significant in shaping the flow of discourse. We situate our findings within the broader context of competing political narratives leading up to the 2016 presidential election. This work provides insight into the underlying structure of politicized discourse on Twitter, and demonstrates the utility of a shared audience network relation for guiding interpretivist analysis of Twitter conversations.

The addition of cholesterol to pure phospholipid bilayers changes the structural properties of the composite membrane. A well-known example of this change is the condensing effect: the average thickness of the bilayer increases, and the average area per molecule decreases, upon addition of cholesterol. We study the spatial extent and variation of these structural changes. We use all-atom Molecular Dynamics computer simulations of mixed cholesterol/dipalmitoylphosphatidylcholine (DPPC) bilayers to quantify lipid structure in a spatially resolved manner. We use order parameters such as area per lipid, tail ordering, and head group tilt angle to measure the ordering potency of a single cholesterol molecule. One challenge when working on a molecular scale is to define a unique area for each lipid in the bilayer, which we have achieved using the Voronoi tesselations of the leaflets. We show that significant changes in phospholipid structure can be observed within the first two solvation shells of the sterol molecule. Our findings provide a simple and direct way to test the controversial Umbrella Model, describing phospholipid-cholesterol interactions and the structure of the bilayer, to improve and expand the theory.

Direct verifications of the validity of Newton's equations for vacuum free-fall in introductory physics laboratories are notoriously difficult due to the presence of atmospheric interactions between falling objects and their environments. The theoretical result, for example, that the gravitational acceleration of an object is independent of its mass is rather easy for observant students to experimentally reject under normal laboratory conditions. What is not so easy is to experimentally verify the validity of mass-independence in-vacuo. This is hindered by inaccessibility to vacuum environments or access to one small enough to fit in a classroom; such access is limited to organizations like NASA. We present here the construction and testing of a portable 1.5m tall extruded-acrylic vacuum free-fall apparatus designed for repeated-measures experiments of time-of-fall with a 2.5g, 40mm diameter magnetic ball over atmospheric pressure ranges 10-4 Torr < p_chamber < 760 Torr. The apparatus includes 12V electromagnet catch and release system within its all-externally-housed actuating mechanisms, automatic digital data acquisition, and a planar optical laser detection system. By comparing time trials in atmosphere and in vacuum, the difference in free-fall times will be isolated and found to be due to air resistance which is not accounted for in the original Newtonian formula. Parameters for this apparatus can be modified to test other ideas for fluid dynamics, such as coefficients of drag and the effects of fluid density on objects as they move through a fluid while using the already present vacuum environment as a control.

The meeting of the fresh water from the Snohomish River and the salt water from Possession Sound creates a unique habitat that is important for many species that make up the biodiversity within this ecosystem. The purpose of this investigation was to observe seasonal variation trends from the pH, chlorophyll, dissolved oxygen (DO) and temperature data at the Deep site in the Possession Sound in Everett, Washington. The study was conducted using a Yellow Springs Instrument 85 and Exo Sonde to collect temperature, dissolved oxygen, pH, and chlorophyll data between 0-3 meters in comparison to 7-10 meters in the the water column. It was hypothesized that during the spring and summer months, there will be a higher chlorophyll concentration due to photosynthesis causing an increase in pH and DO throughout the water column. We also predicted during fall and winter months, there will be a decrease in photosynthesis causing a decrease in chlorophyll, pH, and DO. The hypothesis was not supported. There was an inverse relationship between temperature and dissolved oxygen; as temperature was increasing during the summer months, dissolved oxygen was decreasing. When the temperature was decreasing during the Fall and Winter months, there was a increase in dissolved oxygen.

Post-surgical endophthalmitis is a serious vision-threatening inflammation following any ocular procedure. Detection of endophthalmitis with culture has been traditionally used; however, endophthalmitis caused by viruses are overlooked. In a 1995 Endophthalmitis Vitrectomy Study by RK Forster, 33% of patient samples were culture negative. This represents a significant portion of patients who could have had endophthalmitis due to viruses. Using alternative techniques, we can characterize viral pathogens found in post-surgical endophthalmitis with qPCR (quantitative polymerase chain reaction) and polymerase chain reaction (PCR). qPCR quantifies the number of viral DNA strands whereas PCR indicates presence or absence of the viral DNA. This allows doctors to treat more patients which would have been previously thought to not have endophthalmitis. Patients diagnosed with endophthalmitis following intraocular procedures were recruited from Wills Eye Hospital and University of Washington from Sept 1st , 2014 to Dec 31st , 2015. All patients underwent vitreous or aqueous tap. The specimens were sent for microbial culture. Following DNA extraction, qPCR for Torque Teno Virus (TTV), Merkel Cell Polyomavirus (MCV), and actin were performed in all samples. qPCR results were verified via PCR and gel electrophoresis. A total of 29 samples were collected (12 culture-positive [41.38%], 17 culture-negative [58.62%]). Overall, 18 (62.07%) samples were positive for TTV and 15 (51.72%) for MCV. The presence of TTV or MCV was not associated with culture positivity by Fisher’s exact test. Viral DNA sequences are unexpectedly common in endophthalmitis. The role of these viruses in this disease process remains unclear.

Supernovae enrich the interstellar medium, allowing it to create fascinating and exciting objects such as stars, planets, and people. However, supernovae are difficult to study because, most of the time, we do not observe these supernovae directly, since the whole event can occur in a very short time. Fortunately, they leave behind bright remnants that can give us information about the supernova and even the stars that existed before. We have observed one such supernova remnant in the nearby spiral galaxy NGC 300, which is approximately three million light years away from Earth. We have acquired multi-wavelength data of this source from Chandra X-ray Observatory, XMM-Newton X-ray Observatory, and Hubble Space Telescope. Using this variety of telescopes, we can understand how this supernova remnant behaves in both the x-ray and optical energy bands which gives us a more complete picture of our source. From our Hubble data, we gain detailed measurements of the size and shape of the source. From our Chandra and XMM-Newton data, we can shed light on the properties of this supernova remnant by performing a spectral analysis on the x-ray data. This spectral analysis gives us information about the temperature and metallicity of the source, as well as the density of the surrounding gas. With these quantities, we can better understand this supernova remnant and supernova remnants like it which, in turn, let us better understand the life of the star that created this source.