McNair Scholars

Poster Presentation Abstracts & Schedule

Day 2 – Tuesday, April 19

1 – 3 pm: Alder Hall Commons

All posters will be showcased at the same time during this 2-hour session. Posters are grouped in small clusters according to subject matter.

Education, Humanities & Legal Studies

1. Why Do Youth Programs Fail?

Presenter:

Thomas Nguyen, Suffolk University

Mentor:

Debra Harkins

Abstract:

Much of youth programming demonstrates long-term emotional, cognitive, and social benefits. Unfortunately, research finds that many youth programs fail. This study seeks to better understand why some primary and middle school youth programs in the Boston area fail to meet their mission. Looking into public records and anecdotal experience, I investigated programs that have closed or converted their programming. We found that programs fail or convert due to youth lack of interest and lack of funding. Through each program’s IRS tax forms and Annual Report, along with news articles about said program, I have investigated which programs succeeded and which programs failed. Program failure was defined as a youth program that shut down entirely or had its purpose or target audience changed. This may include a youth program originally meant for elementary and middle school students to a high school program in which the youth are either receiving a stipend or being paid by another program to participate in the current one. Failure may be caused by a lack of funding, lack of interest or uncertainty of the program’s future. The result of this research saw the components needed for a successful youth program are unique and interesting approaches to a topic or issue, established connections with and funding from a university, school district or company (philanthropy), empathetic and diverse staff members. Findings help inform youth programs to be more effective with their funding or staffing and how to avoid potential issues in the future.

2. An Investigation Into the Motivations Behind Black Undergraduates

Presenter:

Sumaya Mohammed, University of North Texas

Mentor:

Dr. Karisma Morton

Abstract:

In this case study, we aim to increase our understanding of the reasons Black undergraduates choose to pursue an education degree as well as their educational experiences leading up to their decision. The participants include Black students enrolled in a teacher preparatory program at a minority-serving institution in the Southwest. Through surveys and interviews with the participants we find that Black undergraduates’ decisions to pursue education degrees are based on a variety of factors, both intrinsic and extrinsic. Findings from this study have greater implications for the recruitment and retention of Black preservice teachers.

3. Deserting Just Deserts: Contextual Blindspots in the Retributivist Theory of Punishment

Presenter:

Jesus A. Raya, California Lutheran University

Mentor:

Dr. Brian Collins

Abstract:

The retributive theory of punishment contends that individuals who commit crimes deserve a punishment proportional to the gravity of their crime. In this paper, I offer a critique of retributivism specifically aimed at its principle of desert due to its failure to consider mitigating circumstances (i.e., justifications or defenses) based on the background of the individual. I will contend that two specific blindspots are especially egregious – socioeconomic status and race, where these particular circumstances lead to deprivation and adversity and seem to be directly tied to the crime committed. This critique draws from and expands on Judge Blazelon and Richard Delgado’s arguments in favor of the “Rotten Social Background” defense. This defense holds that severe economic deprivation ought to be recognized as a legitimate legal defense capable of justifying an acquittal or lesser charge in relevant cases. Ultimately my argument leaves readers to decide whether they want to abandon retributivism because of these problematic contextual blindspots (that is, the failure of retributivism to consider socioeconomic and/or racial backgrounds), or continue to endorse retributivism with the problematic acknowledgement that there is a connection between deprivation and crime. This second option is problematic because it carries the implication that people are deserving of their punishment because they are also deserving of their minoritized, oppressed, and deprived backgrounds. This would be to bite the bullet on classicism, racism, and sexism – I assume that most people are uncomfortable taking a bite of that desert.

4. Barriers in Immigration Court: The Nexus Between Language Access and Due Process

Presenter:

Abi Heath, University of Washington

Mentor:

Dr. Carolyn Pinedo-Turnovsky

Abstract:

Limited English proficiency (LEP) is a significant challenge for persons appearing in the United States immigration court system. The courts often fail to provide adequate language services that address the problems of interpretation. The aim of my project is to investigate the nexus between language accessibility and due process by seeking to answer, “To what extent has the Fifth Amendment of the United States Constitution been an effective framework for lawyers to promote meaningful legal access to LEP individuals?” The Fifth Amendment of the United States Constitution provides that no person should be “be deprived of life, liberty, or property, without due process of law.” Without adequate language services, LEP persons cannot meaningfully participate in court proceedings, and are subsequently denied their entitlement to due process. Little attention has been given to this structural obstruction of justice revealing the need for a study about power and language in the immigration courtroom. To answer my research question, I analyze court decisions in which lawyers have utilized the Fifth Amendment to advocate for clients in Immigration Court and assess the success of those arguments. Results could inform whether judges are likely to interpret the Fifth Amendment as a protection of interests for LEP individuals in immigration court proceedings. With these findings, I hope to inform and assist legal advocates in improving accessibility essential to meeting justice in immigration court proceedings.

Social & Behavioral Sciences

5. Food Expenditure Patterns of Mexican and Hispanic Consumers, A Cross-Country Comparison

Presenter:

Tyson D. Ramirez, University of North Texas

Mentor:

Dr. David Molina

Abstract:

This paper will examine food expenditure habits of Mexican citizens and Hispanic-Americans in an attempt to construct a model that may predict future expenditure habits of culturally similar populations across borders. We will concentrate on observing food expenditure trends in Mexican consumers and the potential relationship to food expenditure trends in Hispanic-American consumers, as well as changes over time in these trends. While absolute and relative income groups may affect expenditure levels, we anticipate that there will be a strong correlation between the food expenditure habits of the two populations. Further analysis will reveal the extent to which (if at all) the expenditure patterns of the two groups correlate with each other. Further analysis may also reveal what other factors may have a significant influence on food item expenditure amongst these populations.

6. Disability and Accessibility

Presenter:

Caroline I. Miles, Suffolk University

Mentor:

Dr. Abraham Peña

Abstract:

The ADA (Americans with Disability) legislation, passed in 1990, was the first of its time to provide basic rights to individuals with disabilities. Although a judicial success for this marginalized community, this population still experiences societal dissociation based on their perceived status in society. This study explores the personal and anecdotal evidence of lingering societal stigma faced by many disabled individuals in the Southeastern United States. The two main forms of disability analyzed were apparent (physical) and silent (invisible) disabilities. From these two groups, findings suggest individuals with disabilities report lower levels of happiness, and confidence than their abled-bodied counterparts. To accommodate the expected mental and physical duress that disabled people will go through in their life, it is imperative that these people have equal access and accessibility to healthcare, prosperity, and avenues in order to be socially integrated into society.

7. Understanding Academic Adversity: Familial Stressors and Academic Achievement Amongst Various Undergraduate Racial Minority Students

Presenter:

Zainab K. Babalola, University of Texas at Austin

Mentor:

Dr. Tia Madkins

Abstract:

Within higher education, a likely consequence of being a racial minority is having lower economic, social, and cultural capital. When coupled with distinct shifts in the traditional family structure this may result in decreased academic well-being in students. In this study, I seek to discover if a relationship exists between the academic achievement of racial-ethnic minority, first-generation, and low-income students and adverse familial experiences. More specifically, I seek to unpack the ways adverse familial issues may compound with students’ social identity markers, such as racial minority status, to ultimately impact students’ academic achievement. I measure academic achievement through The University of Texas at Austin’s grade point average (GPA) scale. I use students’ reported racial minority status, first-generation status, socioeconomic status, familial issues, and academic achievement to determine the extent of adversity they face. Participants include 7 students who are currently enrolled at UT. Data sources include surveys and interviews followed by an analysis with elemental methods such as In Vivo, descriptive, emotion and inductive/deductive coding. Along with second cycle pattern analysis, this method ensures that the conclusions drawn are a direct result of the data collected. I hypothesized that a pattern of adverse familial issues coupled with first-generation status and low socioeconomic status would negatively impact academic achievement. I concluded with themes that were drawn from the categories of data that demonstrated that it is unlikely to be able to draw an absolute conclusion on whether having adverse familial issues compounds with being a minoritized student to impact academic achievement. This is due to the many environmental factors at play, including the COVID-19 pandemic. To gain a more accurate understanding of the relationship between being a minority student and academic achievement, I would like to conduct a long-term study to interview students each year of their academic journey.

8. How COVID-19 Cleanliness Measures Has Affected Ridership of the MTA

Presenter:

Savanna Nolau, Suffolk University

Mentor:

Dr. Jarvis Chen

Abstract:

Since the start of the pandemic in March of 2020, New York City has changed their regulations regarding operations of the Mass Transit Authority. To make passengers and staff feel safe while using the subway and bus system, the cleanliness protocols were heightened, and various new protocols were added. This study surveys real riders of the MTA during the summer of 2021 to gage how the city’s new measures are affecting ridership. In addition to survey data, there was also observational data collected on the claims made by the MTA on how they are handling guidelines and cleaning stations. The results of this study are aimed to help inform MTA officials of their successes, failures and what they can do to promote ridership using the responses of riders.

9. Child Abuse During COVID-19

Presenter:

Urania Mendoza, University of Washington

Mentor:

Debi Talukdar

Abstract:

One in seven children will experience child abuse in their lifetime. While places like schools and day care facilities were once able to detect signs of abuse, the move to online school and closing down of childcare facilities have made it more difficult for children to seek out help during the COVID 19 pandemic. I review current papers that investigate how cases of child abuse have been reported over the past two years and ask the question: How has the COVID-19 pandemic impacted the reporting of child abuse cases? I found readings from the U.S, France, Uganda, and Canada that included keywords such as “child abuse”, “COVID” and “underreporting”. Findings include some countries finding higher rates of reporting which may be accounted for due to the widespread domestic violence campaign that brought awareness to, other countries reporting lower cases which may be accounted for due to the lack of contact with education personnel who are critical reporters, and the potential for online predators to seek out children who are mostly online all day. Overall, this research is important for childcare providers, teachers, and social workers who have been working during the pandemic to provide resources for their students who they may not be seeing in person and to further a conversation of continuing research on the impact of remote learning on child welfare.

10. Motivations for a Green and Just World: Analyzing Narratives of Young Environmental Activists in the Pacific Northwest

Presenter:

Ariel M. Segura, Portland State University

Mentor:

Dr. Amy Lubitow

Abstract:

The United Nations’ Intergovernmental Panel on Climate Change’s most recent 2022 report noted significant concern regarding the state of our environment. Research shows youth-led activism is beneficial to society for a number of reasons, such as their ability to aid in shaping policy and practice. Despite the interrelated series of environmental disruptions, a need for positive environmentally-conscious behaviors and actions, and a growing number of social movement organizations, there is limited engagement from the public regarding environmental activism. Thus, the purpose of this research is to better understand the motivating factors for youth participation in environmental activism. Through the lens of various justice-oriented frameworks, I plan to interview fifteen young people between the ages of fourteen and 21 engaged in the environmental movement. The individual audio-recorded interviews will include open and close ended questions and will be analyzed using a qualitative data analysis program. Youth participants will be recruited through a combination of in person and online engagement with schools, youth-serving organizations, and community groups. I am collecting data and predict there will be some common themes among participant narratives that reveal important insights about young people’s motivations for activism. I believe these findings have the potential to highlight the role young people can hold in shaping the future of our environment, in addition to the relevance of environmental education in K-12 curriculum. Additionally, local governing bodies, and community organizations can use this study to strengthen the argument for the centering of young people’s voice and participation.

11. Investigating Language Interactions Among Mexican Immigrant Mother-Child Dyads

Presenter:

Lauren Morales, University of Washington

Mentor:

Dr. Amy Pace

Abstract:

According to the Linguistic Society of America, children acquire language through interaction with parents and others who are in their environment. Studies also have shown that the types of questions that parents ask can be important for language development. The purpose of this study is to examine the language and communication between Mexican immigrant mothers interacting with their children through a diverse set of play-based activities. To accomplish this, I examine audio transcriptions of play interactions in 12 mother-child dyads who speak primarily Spanish at home. I analyze the audio transcripts, look for communication patterns and the type of questions mothers ask, and categorize the types of questions mothers ask to see if they are consistent across all the audio transcripts. When mothers ask open-ended or complex questions (e.g., what will happen next? Why do you think she said that?) children have the opportunity to build language and thinking skills. As part of this research, I am coding the transcripts to identify the types of questions mothers ask. Specifically, I am interested in how mothers use WH-Questions (e.g., who/what/ when/where/why/how) in Spanish (e.g.,
quién/qué/cuándo/dónde/porqué/cómo). By analyzing mothers’ questions, I can learn about the kinds of language input that supports child language development, as well as inform and expand understanding about how to enhance language acquisition in children from bilingual contexts.

12. People, Parks, and the Pandemic: How Public Green Spaces Shaped Human Wellbeing During
COVID-19

Presenter:

Katelyn McVay, University of Wisconsin – Madison

Mentor:

Dr. Samuel Dennis

Abstract:

The COVID-19 pandemic has reshaped how we use green spaces and how we perceive the health benefits of visiting them. With limited access to indoor gathering spaces due to pandemic restrictions, studies show significant increases in visits to public parks. Spending time at the park has been shown to elevate the perceived mental and physical health of the user. If more people have been visiting green spaces, could the health benefits from these visits potentially outweigh the negative health effects attributed to pandemic-related stress? To investigate this, random intercept interviews were performed at the University of Wisconsin Arboretum. This study uses a mixed-methods approach to survey techniques and data collection with the primary forms of analysis being Likert scales, yes/ no questions, and open-ended responses. Data collection occurred in Madison, WI which may limit how the findings can be applied to more urban and rural populations. Results show that subjects have experienced significant health benefits from visiting public parks during the pandemic, and these visits have decreased pandemic-related life stressors overall. This study will further our understanding of how the outdoors can be used as a coping mechanism and why visits to the outdoors during stressful life events are necessary.

13. Effects of Early Language Environments on Metaphorical Comprehension

Presenter:

Wendy V. Garcia, University of Washington

Mentors:

Dr. Ariel Starr and Dr. Taylor Petersen

Abstract:

There is often a stigma associated with bilingualism due to the mistaken belief that bilingual children develop language skills at a significantly slower rate than monolingual children. To combat this misconception, it is necessary to conduct research showing that bilingualism either helps or does not harm language acquisition when compared to monolingualism. This study attempts to meet that general need by determining how bilingual and monolingual children compare in metaphor comprehension. We decided to focus on metaphors because they can be challenging for children to interpret correctly, and we hypothesized that bilingual children would perform better on metaphor comprehension tasks than monolingual children because of their regular use of second labels. We investigated how 127 monolingual and bilingual children ages 2.5-4.5 years old responded to a set of metaphors by using a game-like format to gather data via Qualtrics. Specifically, we gave children a series of 10 metaphor questions in which they needed to choose which of two pictures best fit a provided metaphorical phrase. Children were also given 10 vocabulary questions derived from the metaphor questions to ensure that children knew the literal meaning of the words. Results showed that bilingual children performed slightly but significantly better than monolingual children on metaphor comprehension questions. Currently, we are also collecting data for a follow-up study to replicate and extend our findings. These results are important because they give us insight into how child language development differs between monolingual and bilingual children and counter negative stereotypes about bilingualism.

Psychology

14. Role Network Analysis to Inform the Design of a Clinical Deterioration Response System for Cancer Patients

Presenter:

Eve Vazquez, University of Central Florida

Mentors:

Dr. Shilo Anders and Dr. Megan Salwei

Abstract:

Cancer patients face unexpected, yet preventable, risk of clinical deterioration during treatment. Limited patient-clinician communication makes it challenging to identify at risk patients for early intervention prior to deterioration; however, clinical decision support (CDS) tools may offer a way to support this early identification. My research sought to identify the roles, activities and tools involved in outpatient cancer care to inform the design of a CDS tool predicting risk of clinical deterioration. Researchers conducted 4 semi-structured interviews with clinicians at a head and neck cancer clinic and completed 11 clinic observations. Each interview was audio-recorded, transcribed, and uploaded to a qualitative data analysis software, Dedoose, along with the observation notes. I coded the data for roles, individual and team activities, group meetings, and tools used. The results were used to develop a role network that visually displayed the data, including the identification of 24 different roles involved in a patient’s cancer care, including physicians, schedulers, nurses and lay caregivers. The results also found approximately 100 activities that revolved around a patient’s care network, and numerous tools including the electronic health record, patient portal, phone, and paper documents. The role network provided insight into the complexity of patient care, and the high-level of communication that occurs within the cancer patient care system. This research serves as a foundation in which to analyze additional cancer clinics and connections among clinics for optimal patient care and further CDS implementation.

15. The Power of Affirming Spaces for QTBIPOC: An Intersectional Analysis

Presenter:

Joshua M. Hanock, University of Illinois at Urbana-Champaign

Mentor:

Dr. Ghassan Moussawi

Abstract:

Community is an essential need for humanity and when Queer Trans Black Indigenous People of Color (QTBIPOC) are lacking this need, the effects of its discovery may prove powerful for individuals accustomed to alienation and discrimination. Existing literature has reviewed why community among QTBIPOC is vital, citing possible heteronormativity and cisnormativity experienced by communities of origin compounded by racism and cisnormativity experienced by the dominant white lesbian, gay, bisexual, transgender, questioning plus (LGBTQ+) communities. Existing literature has also explored arenas for community finding such as rallies and protests. Yet, the literature on what “community discovery” means for the collective humanity of this population is less than thorough. Through an analysis of the existing literature, I will explore the human impact that community discovery has on QTBIPOC. Guided by the theoretical lens of intersectionality and multiple minority stress theory my aim for this research is to highlight the unique circumstances curtailing and empowering the discovery of community for QTBIPOC and its healing and liberating significance. Such findings can aid academics in remembering to consider the impacts that racialized marginalizations have when researching LGBTQ+ experiences in the US.

16. The Effects of Inpatient Psychiatric Socio-Physical Experience on Posthospitalization Treatment

Presenter:

Yasmin Landa, University of Washington

Mentor:

Dr. Heather Evans

Abstract:

The deinstitutionalization of psychiatric hospitals in the 1970s and 1980s improved the physical and social environment of psychiatric hospitals and aided in the transition away from prison-like hospitals. Still, there remains opportunities for growth if the psychiatry field aims to engage and provide patients with a fully therapeutic setting. This research explores the relationship between a patient’s social and physical experience in a psychiatric hospital and their engagement with mental health treatment after discharge. I analyze the physical environment of a psychiatric unit and explore mental health professionals’ views on guidelines for maintaining a therapeutic environment in these spaces. Using qualitative methodology–ethnographic observations of psychiatric hospitals in the Seattle area, photographs, and interviews–I discuss the environmental factors of psychiatric rooms through the evaluation of room design and its effectiveness in creating a therapeutic environment. Preliminary findings point to a possible relationship between consent and a patient’s receptivity to post-hospitalization treatment after being exposed to the psychiatric hospitalization environment. Additional results could indicate a negative relationship between a patient’s socio-physical experience in an inpatient psychiatric facility and their engagement in future treatment after discharge. These findings will help providers improve upon these experiences and increase a patient’s receptivity to post-hospitalization treatment.

17. Impact of the Strong Black Woman Schema on Workplace and Education Among Black College Women

Presenter:

Alla Jones, University of North Texas

Mentors:

Dr. Martinque Jones and Gabriella Gaskin-Cole

Abstract:

The strong Black woman (SBW) schema is a widely accepted gendered racial archetype of Black womanhood (Jones et al., 2021; Nelson et al., 2016; Woods- Giscombé, 2010). Young Black women who ascribe to the SBW schema report psychosocial benefits (e.g., self-esteem, Woods-Giscombe, 2010; Watson & Hunter, 2016), as well as many more consequences (e.g., depression, stress, and even suicidality, Castelin & White, 2022; Donovan & West, 2015; Jones et al., 2021; Watson & Hunter, 2015). Although the psychological consequences of identifying with the SBW schema have been established, far fewer studies have examined how the SBW schema impacts Black women in the workplace and educational contexts. To address this gap in the literature, the current study drew upon data extracted from a larger qualitative study of strength-related socialization messages and their impact on Black college women. In this study, the researchers conducted semi-structured interviews with 15 Black college women (Mage = 19.73; SD = 1.16) and will analyze relevant data for this study using consensual qualitative research (CQR; Hill et al., 1997) approach. Findings derived from this study have the potential to broaden our understanding of the impacts of the SBW schema on Black women across contexts.

18. Maternal Stress in Pregnancy and Infant Cognitive and Physical Development

Presenter:

Jeniffer Kyule, University of Washington

Mentors:

Dr. Liliana Lengua and Michele Smith

Abstract:

According to the World Health Organization, 10% of pregnant women and 13% of recent mothers suffer from adverse mental health, particularly depression. The pregnancy process is a tumultuous experience as there are many physical and psychological changes, as well as other prenatal stressors like limited resources that disproportionately affect low-income mothers negatively impact maternal mental health, birth outcomes, and an infant’s cognitive and emotional development. The study examines the impact of maternal mental health on low-income mothers and children and how mindfulness can be a tool of intervention. I conducted a literature review using preliminary outcomes of ongoing, longitudinal research examining maternal well-being following a mindfulness-based intervention among low-income new mothers with data acquired from the New Moms Connect study. Findings from the NEW Moms Connect Study are preliminary and not yet available. However, similar research (Norona-Zhou et al., 2022) showed that, compared to treatment-as-usual, mindfulness-based intervention for expectant mothers was related to advantageous physiological and behavioral reactivity and regulation patterns for their 6-month-old infants. Infants in the treatment-as-usual group had slower sympathetic activation and recovery in a high stress behavioral paradigm, while those in those in the intervention group had greater proportions of self-regulatory behavior. Research is still ongoing regarding the long-term effects of mindfulness during the perinatal period. This research is important as it provides data on ways to improve maternal health of low-income mothers who have been grossly underrepresented in previous similar research. These findings also indicate the potential benefits of mindfulness programs geared towards serving low-income communities.

19. Attitudes and Engagement: The Use of AI Feedback and an Online Office for Treating ADHD Patients

Presenter:

Jessica M. Monroy, University of Washington

Mentor:

Dr. Margaret Sibley

Abstract:

STAND (Supporting Teens Autonomy Daily) is an evidence-based practice modular therapy that helps ADHD teens build academic and interpersonal skills. STAND has been shown to be cost-effective in promoting evidence-based practices compared to the usual-care psychological services and medication for ADHD. The purpose of this study is to assess the utilization and efficacy of online resources and training tools for therapists and supervisors implementing STAND. Six therapists and four supervisors administering STAND use LYSSN, an AI-based tool that provides feedback on therapy audio tapes, and Care4, a dashboard that provides supervisors with a library of resources. Therapists login onto the Care4 system after a therapy session and answer questions about feedback they received from LYSSN and the preparation for their session. Supervisors could use the data collected from the therapist’s LYSSN score and formulate their feedback. There was an average time of 11 to 30 minutes that therapists spend reading through the STAND session description and preparing client worksheets. An average of 1 to 10 minutes reviewing LYSSN feedback, watching STAND demonstration videos, seeking consultation from the STAND research team, and reviewing feedback from their supervisor. When it came to finding LYSSN feedback to be accurate and credible, therapists found LYSSN to be occasionally accurate and credible. Using AI feedback, the workload of supervisors and therapists could be reduced and give feedback more efficiently to them with regards to STAND treatment and other evidence-based treatments.

20. Differences in Symptoms Following Sexual Assault in Bisexual and Lesbian Women

Presenter:

Abril Beretta, University of Washington

Mentor:

Dr. Emily Dworkin

Abstract:

Sexual minority women (SMW)—bisexual women in particular—have greater risk of experiencing sexual assault and other types of traumas and stress, as well as heightened symptom severity than heterosexual women. Bisexual women in comparison to lesbian women have shown a higher risk of reporting worse mental health. To our knowledge, only three studies have examined differences in mental health in bisexual and lesbian women who have been sexually assaulted, but their results were contradictory. As such, the purpose of this study is to investigate the symptom differences between bisexual and lesbian women following sexual assault to better understand their differences. To do this, I conducted a secondary analysis of an existing national data set that surveyed 1057 SMW. I used bivariate statistics tests to compare lesbian and bisexual women regarding their mental health following sexual assault with a focus on anxiety, depression, and PTSD symptoms. No statistically significant differences were found between the two groups as a function of sexual assault, but bisexual women reported greater symptoms of depression regardless of sexual assault status in comparison to lesbian women. A potential explanation for this difference could be that bisexual women may have higher levels of minority stress (i.e., stress resulting from societal prejudice and other stigmatizing experiences) than lesbian women, which has been connected to heightened mental health issues. Future studies should consider minority stress as a factor when studying the two sexual identities to better tailor research and treatment to bisexual and lesbian women’s unique identities.

Environmental Science & Ecology

21. An Assessment of the Potential for Arbuscular Fungi to Aid in Corn Uptake and Availability of Iron

Presenter:

Caitlin R. Costello, Portland State University

Mentor:

Dr. Jen Morse

Abstract:

Iron deficiency is a major contributing factor to global malnutrition, illness, and death. According to the Food and Nutrition Bulletin, this deficiency accounts for 841,000 deaths and approximately 35,057,000 “disability-adjusted life years lost” annually. Malnutrition rates can already be attributed to rising emissions and global temperatures. As climate change advances, these effects will only be exacerbated. One way to combat this is to maintain or increase the iron uptake of the globe’s major crops, such as corn. Research has shown that as CO2 emissions elevate, the macronutrient content in crops decreases. My research assesses the potential for arbuscular mycorrhizal fungi, which is known to aid in plant uptake of macronutrients, to mitigate the climate change effects on the nutritional content of crops by increasing the availability of usable iron in the soil. I am using a short harvest variety of corn and growing a control group as well as a group inoculated with an arbuscular fungi root dip. Both are grown in a greenhouse setting to limit confounding factors. At the time of harvest, the fruit is dried and an acid digestion process performed. Using infrared assessment, the iron content of each corn group is ascertained. Expected results include the inoculated crops showing significantly increased iron content. If this hypothesis is upheld, this research could show potential options for mitigating the reduction of crop macronutrients without increasing the use of synthetic fertilizers which further global climate change and increase emissions.

22. Species Richness & Floristic Quality Assessments Differ in Their Sensitivity to Lonicera maackii Invasions with Increasing Area

Presenters:

Eduardo Tovar, University of Illinois at Urbana-Champaign

Mentor:

Dr. Jeffrey Matthews

Abstract:

Plant invasions are generally associated with decreased biodiversity and ecosystem function, but the impacts of plant invasions on biodiversity is often dependent on the scale at which a study is conducted. Metrics of community structure and quality differ in their sensitivity to invasion with area, resulting in a need for more sensitive, reliable mechanisms of quantifying invasion impact. Lonicera maackii is a shrub native to Asia and invasive across the eastern United States. The impact of L. maackii invasion on species richness and two floristic quality (FQA) metrics, floristic quality index (FQI) and mean coefficient of conservatism (mean C) was studied across multiple spatial extents to determine the spatial breadth of the sensitivity of these metrics to invasion. We selected 12 forest plots: 6 uninvaded plots (<1% L. maackii cover) and 6 invaded plots (>70% L. maackii cover), which we divided into subplots of 1 m², 62.5 m², 125 m², 250 m², and 500 m². Calculated within each plot were the average values for plant richness, mean conservatism, and floristic quality for each spatial extent. Plots invaded by L. maackii displayed lower total richness at smaller spatial scales, with richness within L. maackii-invaded plots recovering to levels found among uninvaded plots at an extent of 62.5 m². Invasion impacted both mean conservatism and floristic quality index by lowering their values in invaded plots at spatial extents equal to and under 125 m2. Consequently, FQA metrics may be more sensitive predictors of the impacts of plant invasions than species richness.

23. The Impact of Invasive Plant and Macroinvertebrate Species on Water Quality of Streams in the Pacific Northwest

Presenter:

Michelle Hesek, Portland State University

Mentor:

Dr. Patrick Edwards

Abstract:

Invasive species are destructive to the health of riparian areas and the water quality of a stream, while also affecting our way of living through undrinkable water sources, decreased biodiversity, and reduced infrastructure stability. Environmental agencies alongside the US Forest Service annually spend billions of dollars towards its management. Water quality is especially decreasing within streams of the Portland Metro area, and a major aspect in evaluating its health is through a biological assessment. This research aims to find how invasive plant species affect riparian and stream quality through the index of biologic integrity. The study area will include 7-8 streams in the Portland Metro area, with the inclusion of macroinvertebrate data collected by Patrick Edwards and his community scientists. Using the Community Science- Indices of Biologic Integrity (CS-IBI) scale, we can evaluate water quality from the field sampling results. Field data will be collected using a quadrant and systematic random sampling methods during the summer months, and evaluated in excel. Based on relevant research on water quality and invasive species, it is implied that there will be a negative relationship between greater invasive species cover and the CS-IBI score. These results will add to the recent studies done on this topic, alongside easier ways to reduce invasive species and suggest improvements to water quality in marginalized communities.

24. The Angry Basking and Over Stacking of Freshwater Turtles

Presenter:

Camoya Evans, University of Wisconsin – La Crosse

Mentor:

Dr. Roger Haro

Abstract:

The proposed research is to help understand the ways that human activity on the Mississippi River has shifted the behavior of freshwater turtles. This research is focused on the presence of deadwood in the river, and the ways that turtles interact with the deadwood. The proposed question is whether turtles are more aggressive with one another over the few logs present in the Mississippi River in human-dominated areas along the Mississippi River. We answer this question by observing three locations around the River from June to August. We collected data based on their over stacking and interactions with one another and the down wood in the water.

Biomedical Sciences, Bioengineering & Biophysics

25. Optimizing Computationally-Designed Fluorescence-Activating Protein with DFHBI

Presenter:

Justin P. Nguyen, Wesleyan University

Mentor:

Dr. Colin Smith

Abstract:

Designing proteins from scratch that have novel, three-dimensional folds has become well-established in the field of protein design and engineering. However, a more difficult problem to tackle in this field is incorporating functionality with a practical framework in mind. In the Smith Lab, we investigate the relationship of protein’s structure and how that may affect its function using computational simulation and modeling methods and nuclear magnetic resonance (NMR) spectroscopy. In particular, I work with 13 computationally-designed β-barrel proteins termed mini-Fluorescent Activating Proteins (mFAPs), originally created by the Baker Lab in 2018. These synthetic proteins activate an exogenous, GFP-derived chromophore, 3,5-difluoro-4-hydroxybenzylidene-imidazolinone (DFHBI), enabling it to fluoresce ten to hundred-fold more than free DFHBI in solution; it is known that DFHBI fluoresces in a planar-Z conformation. In our computational simulations, we found that the experimentally brighter variant holds the chromophore in a planar-Z conformation for a longer time compared to the experimentally dimmer variant. My project involves the modeling software Rosetta, which allows me to input exhaustive mutations and outputs how stabilizing or destabilizing a mutation is with respect to holding the chromophore in a planar state. Preliminary results show that the prediction algorithms work in some cases in improving brightness, however at its current stages, refinement of the data processing is necessary for accurate predictions. This model protein-ligand system serves as a basis for optimizing the structure and function of de novo proteins and more broadly, expand the necessary tools for optimizing functionality as therapeutics, biosensors, or enzymes.

26. Reengineering Candy to Aid Collection of Oral Bioanalytical Samples

Presenter:

Damielle L. Hieber, University of Washington

Mentors:

Dr. Ashleigh Theberge and Dr. Sanitta Thongpang

Abstract:

The rapid rise of bioanalytical testing has renewed discussions in improving the collection of pathogens through oral sampling. Current diagnostic tests for common respiratory diseases such as strep throat are invasive and uncomfortable, especially for children. Undiagnosed cases of these treatable diseases can lead to serious damage. The Theberge lab has developed a saliva sampling device intended to be a child-friendly alternative to current sampling techniques for home and clinical settings, called CandyCollect. CandyCollect contains sugar free candy in the design to both appeal to children and also act as a built-in sampling timer, which is a unique component that requires additional testing and standardization to be appropriately implemented. Our goal is to modify the existing device with respect to the intersectionality of safety, design, effectiveness, and ease of application by focusing on the candy component. Several types of candy were made using different ingredients, flavors, and textures, and the interaction of these candies with oral bacteria and saliva was investigated. We engineered the device using rapid prototyping, computer-numerical-control (CNC) milling, and silicone mold making. These methods allow for the flexibility to modify the design of the candy and sampling device. The next steps will be testing the new devices in human subject studies for adults to receive usability feedback and adapt the device components accordingly. The findings of this study will offer an improved method for child diagnostics and encourage other industries to redesign traditional sampling procedures.

27. On-Demand Protein Photoactivation Within Hydrogel Biomaterials

Presenter:

Kathy Do, University of Washington

Mentors:

Dr. Cole DeForest and Ryan Francis

Abstract:

Abstract cannot be printed due to privacy restrictions on the work of this lab. Go to the research talk if you want to know more!

Microbiology & Molecular & Cellular Biology

28. Life Will Find a Way: Investigating Entombed Microorganisms in Gypsum at Great Salt Lake as a Model for Studies of Gypsum on Mars

Presenter:

Paulina I. Martinez Koury, Westminster College

Mentor:

Dr. Bonnie Baxter

Abstract:

Modern Great Salt Lake (GSL) and the surrounding Bonneville Salt Flats resulted from the mass evaporation of Lake Bonneville, the largest lake within the Great Basin in the Western United States that formed during the last ice age. Modern GSL is a unique hypersaline environment, hosting a complex community of microbiota that are as distinctive as their host environment in which both halite crystals and gypsum crystals precipitate. Evaporite minerals such as gypsum have also been well mapped on Mars by recent space missions, suggesting the evaporation of large Salt Lake systems. The geology, morphology, and minerals are a few of the striking similarities that render GSL as a potential analogue to evaporative lake regions on Mars. The entombment of halophiles, or “salt-loving” organisms has been demonstrated in ancient to modern minerals. Recent research on entombed microbes in halite have demonstrated microorganisms living on the mineral surface, may contaminate experiments meant to assess the microbes entombed within. I hypothesize that, clay inclusions in GSL gypsum crystals will be enriched with archaea, bacteria, and fungi from the lake’s microbial community. I developed a multi-step surface sterilization protocol that accounts for the unique chemical makeup of gypsum, and a protocol for optimal release of entombed microorganisms within clay inclusions. Preliminary results have provided evidence for the presence of DNA entombed in clay inclusion within Gypsum. This project will generate an optimal surface sterilization of gypsum, the first protocol of its kind for this mineral, as well as the first community-wide genomic isolation and identification of gypsum entombed microorganisms.

29. Microglia-Specific AD Risk Variant BIN1 Deletion in IPSC-Derived Microglia

Presenter:

Mark Lopez Sanchez, California State University of San Marcos

Mentors:

Dr. Nicole Coufal and Dr. Anna Warden

Abstract:

Microglia, the primary immune cell of the brain, plays a critical role in the immune response in neurodegenerative disorders such as Alzheimer’s Disease (AD). There is increasing evidence for the role of noncoding genetic variants in the risk of idiopathic AD, with one of the most associated AD genetic risk variants have been found to reside in a microglia-specific Bridging Integrator 1 (BIN1) gene locus, suggesting that the dysregulation of microglial BIN1 contributes to AD risk. To investigate the role of BIN1, we leveraged hESC and iPSC-derived microglia using CRSIPR/Cas9 to create BIN1traditional loss of function (tKO) microglia cell lines. We evaluated microglia canonical marker expression and morphology using live imaging and immunofluorescence, finding the BIN1 tKO line exhibits a ramified morphology. Additionally, we examined the phagocytic function of the BIN1 tKO line by incorporating phagocytic stimuli such as, fibrillar amyloid-β and Tau. We identified enhanced phagocytic activity with loss of BIN1. To address functional disturbances caused by the deletion of BIN1 in microglia, we performed cell migration assays, lysosome activity assays, and mitochondrial superoxide activity. We determined that the BIN1 tKO line expresses higher basal levels of mitochondrial superoxide and an enhanced effect on the production of superoxide after exposure to stimuli. Altogether this suggests a hyperphagocytic and hyperinflammatory phenotype resulting from loss of microglial BIN1. The long-term goal is to identify microglial pathways involved in AD onset and progression for targeting novel therapies by first understanding how genes regulated by AD risk alleles contribute to microglial function.

30. Isolating the Cell-Invasive Stage of a Cockroach-Infecting Gregarine for Gene Expression Analysis

Presenter:

Mia M. Sanchez, St. Edward’s University

Mentor:

Dr. Daniel Gold

Abstract:

Gregarines are protistan parasites in the phylum Apicomplexa which parasitize the intestinal epithelia of nearly all invertebrate clades via an oral-fecal pathway. Gregarines are most closely related to Cryptosporidium spp., which infects vertebrates and most notably causes cryptosporidiosis in humans. Gregarines’ similar lifestyle and lifecycles to Cryptosporidium spp., allow the largely understudied gregarine parasite to serve as a model for the human infectious parasite. Similar to Cryptosporidium, when gregarine oocysts are ingested, undetermined host cues trigger the release of sporozoites that infect the intestinal epithelia. Despite expressing gene and protein products believed to be associated with host cell invasion, no RNA sequencing of gregarines at the sporozoite stage has been reported. To develop a better understanding of the host-parasite interactions, we plan to develop a method of isolating pure sporozoite samples from a host colony of Madagascar Hissing Cockroaches then sequence the transcriptome of this invasive stage. Our developing method of sporozoite isolation is an in vitro procedure using a combination of different centrifuge speeds and intestinal fluid from the host to induce oocyst excystation. To determine the efficiency of sporozoite isolation, we will examine the supernatant and pellet fractions from the samples via microscopy. Efficiently isolating a pure sporozoite sample will aid in the RNA sequencing process and allow us to shed light on the genes expressed at this understudied invasive stage.

31. Identifying New Roles for the Proteasome Pathway in Congenital Heart Defects

Presenter:

Isabelle Young, University of Washington

Mentor:

Dr. Lisa Maves

Abstract:

Congenital Heart Defects (CHDs) are the most common birth defect within humans and are characterized by having multiple structural problems with the heart, often leading to problems with blood circulation. Our lab has identified a set of genes potentially involved in birth defects that are also expressed during heart development, thus identifying candidate genes for congenital heart defects. Amongst the gene candidates is a gene encoding proteasome maturation protein (POMP). Proteasomes are the main system for protein degradation, but the role of proteasome factors in CHD development is not known. To establish the importance of POMP and the proteasome in proper heart development, I examined zebrafish embryos that are mutant for the POMP gene and the effects of chemical proteasome inhibitors on zebrafish heart development. Zebrafish (Danio rerio) are commonly used to study human development and birth defects because their genes and organs strongly resemble those of humans, and the zebrafish heart develops in only 3-4 days. Using fish that have been tagged with green fluorescent proteins to visualize heart development, I tracked the developing hearts of POMP mutants and proteasome for signs of heart malformation. I expect to find structural similarities between the developing hearts of the POMP mutants and those with inhibited proteasome function, thus indicating the proteasome system as having a prominent role in CHDs. Gaining insight into the molecular mechanisms and genes behind CHDs aids in the development of human patient diagnoses, targeted therapeutic treatments, and an increased understanding of complex diseases.

32. Identifying Combinatorial Therapeutic Strategies for BET Inhibition in Ewing Sarcoma

Presenter:

Rohda A. Yase, University of Washington

Mentors:

Dr. Elizabeth Lawlor and Dr. Shireen Ganapathi

Abstract:

Ewing sarcoma (ES) is a bone and soft tissue tumor that occurs in children and young adults. The tumor is driven by an oncogenic fusion gene that fuses the EWSR1 gene to FLI1, an ETS family transcription factor. The EWS-FLI1 fusion promotes tumorigenesis through transcriptional and epigenetic dysregulation. Despite maximally intensive chemotherapy, the outcomes for metastatic ES patients remains poor, thus the need to identify new therapeutic strategies. Given ES’s epigenetic dependencies, there is strong rationale to investigate epigenetic modifying drugs. Bromodomain and extra terminal domain (BET) proteins function as epigenetic readers that facilitate transcription. I have shown that BET inhibitors (BETi) slow the growth of ES cells in vitro but will not be successful as a single agent. I hypothesize that the combination of BETi with other biologically targeted agents will be synergistic. Based on our preliminary results from RNA-seq data and an in silico drug screen on BETitreated ES cells, we prioritized testing of top predicted small molecule inhibitors. Using standard in vitro cytotoxic assays and calculating synergy using the Chou-Talalay method, my preliminary results showed strong synergy between Copanlisib and BMS-986158. Ongoing studies are testing the selected kinase inhibitors, and promising combinations will be tested in in vivo xenograft models. It is our goal to identify drug combinations that will enhance the cytotoxic effects of BETi in ES.

33. Investigating RIT1 M90I and KRAS G12V at Varying Timepoints in the RAS Pathway

Presenter:

Eric I. Morales-Perez, University of Washington

Mentors:

Dr. Alice Berger and Amanda Riley

Abstract:

Lung cancer is the leading cause of cancer-related deaths worldwide. At the molecular level, lung cancer is caused by altered cellular signaling pathways. One of the most dysregulated pathways is the Receptor Tyrosine Kinase (RTK)/RAS pathway involving many molecular players, including GTPases (proteins that cycle between GTP- and GDP-bound states), which regulates cell growth and proliferation. A gene called KRAS is a well-studied GTPase involved in RTK/RAS signaling, where mutant forms have been shown to increase proliferation and are implicated in cancer. Recently, a gene called RIT1 was discovered that encodes for a GTPase and is also implicated in cancer. However, despite knowing that both mutants are responsible for driving lung adenocarcinoma, its unknown about their differences in terms of expression levels. In this study, given at equivalent levels, I ask, “what are the functional differences and cellular consequences in the MAPK pathway activation overtime in RIT1 and KRAS?” To answer this, I used immortalized tracheobronchial epithelial (AALE) cells treated with doxycycline at various time points to induce V5-tagged KRAS and RIT1 expression. Lysates were collected to measure protein expression levels using the V5 tag for western blot analysis. The hypothesis and results suggest that, given equivalent expression levels of KRAS and RIT1, KRAS would drive more AKT, MEK, S6, and ERK phosphorylation than RIT1. Given the data, although RIT1 is a GTPase, it is a weak form of KRAS. These findings will reveal the functional differences between KRAS and RIT1 when activated in the RAS/MAPK pathway.

Engineering & Computer Science

34. Electronic Mode Stirring Improves Wireless Communications in an Animal Cage Environment

Presenter:

Sara Reyes, University of Washington

Mentor:

Dr. Matt Reynolds

Abstract:

In vivo recordings of neural ensemble activity in non-human primates (NHPs) have contributed to the understanding of how neural activity relates to motor function and intent. Traditionally, neural recording has been conducted in constrained environments, such as head-fixed experiments using bulky wired equipment to achieve high data throughput and reduce measurement noise. To collect more natural data from free-moving NHPs requires a high speed, low power wireless uplink of the neural data. However, wireless communication inside a metal NHP cage suffers from dense multipath interference, due to multiple signal bounces from the cage walls, which decreases communication reliability. We explored an approach called “Electronic Mode Stirring” to obtain better communication reliability. We found that by adding 4 switchable reflecting antennas to the roof of the cage to perform mode-stirring we were able to improve the mean one-way path loss and improved the worst-case signal loss across 126 measured positions in the primate cage. We expect that the reduced signal loss with the mode stirring system will lead to improved wireless communication reliability inside the cage.

35. Designing and Building TinyQuad: A Quadcopter That Weighs 1-2 Grams

Presenter:

Alyssa M. Giedd, University of Washington

Mentors:

Dr. Sawyer Fuller and Dr. Vikram Iyer

Abstract:

Development and testing of sensors and power methods for insect-based robots is a difficult task. Due to the high cost of manufacture with regards to both training time and funds, finding a sustainable and easy-to-produce method to test sensors and power options is essential. Previously, the only option for testing new sensors and power options was using one of the robotic insects, which is risky considering their high costs. Drawing from prior results, we believe a lightweight quadcopter would be faster, easier to produce, more robust, and able to serve as a suitable replacement in sensor testing and development. My goal will be to create the world’s lightest and smallest quad-rotor helicopter, “TinyQuad,” with a target mass of 1–2 g. The new helicopter I have designed will enable the testing of new sensors such as cameras and power options such as radio frequency-based charging. I will demonstrate flight capabilities through utilizing wireless charging, and sensor-based feedback control to improve flight stability and duration. I completed calculations and design of this hardware, and anticipate seeing that the collected flight data supports the utilization of a lightweight quadcopter in insect robotics development. This will allow us to rapidly develop and refine sensors for use onboard the RoboFly robotic insect platform. Creating a working quadcopter would result in accelerated prototyping that allows for more unusual sensor and payload designs, and for further research in developing new sensors and power methods for insect robotics, smaller quadcopters, and improved design of micro aerial vehicles.

36. Effects of Climate Change on the Water System in Koror, Palau

Presenter:

Marley Rabauliman, Washington State University

Mentor:

Dr. Jan Boll

Abstract:

Palau is a tropical island in the South Pacific that experiences effects of climate change due to not only sea level rise, but also the decrease in water security. Half of Palau’s population of over 20,000 people live in Koror, where daily water demand is substantial for households and businesses. During the dry season, Koror regulates water use by having water rationing hours. Climate change is already increasing the dry season length and drought severity. This study investigates how to reduce water rationing hours in Koror. A stock-and-flow systems dynamics model in Vensim was created to simulate Koror’s current water system. Different simulations will evaluate how the water system handles the change in climate patterns, through scenario testing focusing on dry season length and drought severity. Climate patterns in the model will be based on climate data recorded over several years in the areas where the storage reservoirs are located. Model results will show components of the mass balance such as water inputs to the reservoirs, change in storage, and water being pumped out. Tentatively, scenario testing would include the water system during the rainy and dry season and when water hours are in effect. With this information, the model will provide information on ways to improve the system. This goes hand in hand with reducing water rationing hours because improving the system means there will be a sufficient amount of water in the reservoirs to meet demands.

37. Methods Used to Measure the Hydration Kinetics of Cement

Presenter:

Domenica Gachet, Washington State University

Mentors:

Dr. Vikram Yadama and Dr. Manuel Raul Pelaez-Samaniego

Abstract:

Cement hydration determines the properties of cement. It is a process of multiple chemical reactions that take place when water is added to the cement. Chemically bound water can be measured by the degree of hydration. Water bonding produces hydrates that can determine if the cement will prevent corrosion of the reinforcement, increase the strength, and increase the durability of cement. However, the complexity of each chemical stage causes hydration studies to be typically focused on the analysis of pure substances in controlled environments. There are few studies centered on the hydration of cement when mineral admixtures are used. The use of mineral admixtures represents a reduction in the amount of CO2, reduction in costs, and helps recycle by-products from other industries. In this paper I will explore multiple methods currently used to measure the hydration kinetics of cement to determine the best method or combination of methods that can be utilized to obtain accurate results. Some of the methods I will analyze include measurements of heat produced by the specimen (differential thermal analysis, and differential scanning calorimetry), elements reflected in X-rays (X-ray diffraction analysis), and the composition in mass of cement (thermal gravimetric analysis) to measure the hydration kinetics of cement. It is expected that the selected method or methods will help to determine the degree of hydration and the quantity of hydrates produced. The results could potentially be used as a starting point to study the hydration of cement when manufacturing cementitious grout using biochar as an admixture.

38. Determining Kinetic Parameters for a Mathematical Model to Optimize Growth of Cancer-Fighting T Cells in a Perfusion Bioreactor

Presenter:

Zakora Moore, Washington State University

Mentors:

Dr. Bernard Van Wie and Kitana Kaiphanliam

Abstract:

T cell immunotherapy is a highly effective cancer treatment in which the immune system’s inherent ability to fight cancer is amplified by increasing the amount of cytotoxic T cells that are deemed most active within a patient. Current T cell expansion methods are inefficient, resulting in high manufacturing costs, which brings question to the accessibility of T cell therapies. We hypothesize that by producing a mathematical model to predict cell growth, substrate consumption, and metabolite production of a proof-of-concept T cell line over time, we will be able to optimize growth of the cell line in a perfusion bioreactor system. A series of four studies were performed to produce the growth model: (1) measuring yield coefficients of lactate, ammonium ion, and glucose; (2) determining the Monod constant and maximum specific growth rate; (3) finding critical metabolite concentrations above which cells will not grow; and (4) identifying ideal amounts of interleukin-2 (IL-2) that prove most effective supporting cell expansion at low T cell concentrations. Once all parameters are defined, we can confidently configure glucose, lactate, ammonium, and IL-2 levels to apply to the growth model as the T cells grow to high densities in a bioreactor and, as a result, optimize the T cell manufacturing process for cancer immunotherapy treatments.

39. Human and Computer Learning: Same, Similar, or Different?

Presenter:

William Zaudio, University of North Texas

Mentor:

Dr. Michael Thompson

Abstract:

The goal of Artificial Intelligence (AI) is to build artificial systems able to mimic human mental capabilities such as learning, reasoning, and perception. Among these three capabilities, the research will focus on computer learning and cross-compare it with human learning by exploring philosophical discussions of education, deductive learning in computer science, inductive learning in robotics, and non-logicist AI learning through artificial neural networks to provide both; first, a clear understanding of the present potential computers have to learn so as to achieve more effectively the ultimate goal of AI in future years; and secondly, to assess through their similarities and differences points of interaction and complementarity where they can depend on and influence one another through mutual exchange, by finding their appropriate place and use according to their capabilities. From the research, I expect to find that while computer learning may be modeled after human learning, they both are distinct from one another. However, in analyzing their differences, I anticipate to find the contrast between human learning and machine learning is not that broad. Multiple disciplines have evaluated the success of human and computer learning while having different starting points, and it has led to varied conclusions concerning the similarities and differences between them. However, by analyzing these sources, I will provide a much-needed comprehensive contrast of the reach of each while also assessing areas where they might confluence and assist each other’s growth.

40. Independent Measurement Platform for Federated Learning Models on Android Devices

Presenter:

Michael Cho, University of Washington

Mentor:

Dr. Afra Mashhad

Abstract:

Machine learning is a powerful tool that allows us to use data to make predictions and decisions about the world, but it requires expensive centralized hardware and data, is prone to algorithmic biases, and has privacy concerns surrounding the use of required data. In contrast, Federated Learning (FL) allows users to collaboratively train a shared model under a central server while keeping personal data on their devices. FL requires established processes for training and measuring the efficiency of machine learning models on edge devices. This research provides an inclusive framework to federatively train models on Android devices and analyze their computational and energy efficiency. I leveraged a terminal application to install dependencies and natively train machine learning models on mobile devices. Then, I analyzed the device’s efficiency by measuring the computational, energy, and network resources
through terminal applications. This flexible framework can deploy diverse machine learning models for training on Android devices. In preliminary experiments, I used this framework to measure efficiency for a PyTorch obstacle detection model and Tensorflow abnormal heartbeat detection algorithm. Experiments show that training machine learning models on mobile phones efficiently uses CPU, memory, and bandwidth, and it uses minimal energy consumption compared to centralized machine learning. With little to no
examples of FL on Android devices, this framework provides a novel plug-and-play solution for native FL on mobile devices. Research applications will also demonstrate novel methods for using FL techniques to address topics of accessibility, privacy, algorithmic bias, and hardware limitations for machine learning.