Biological Sciences Major
Student Author(s): Trey Zinsmeister, Kaitlyn Snyder, Caitlyn Helton, Lauren Taylor
Faculty Author(s): Dr. Rick Boyce
Author Department(s): Biological Sciences
Community Partner(s): University of Kentucky, Stanford University
Title: Response of Amur honeysuckle to a combination of decapitation and leaf blight
Abstract: Amur honeysuckle is an invasive plant that extensively affects the northern/central Kentucky region. Control is difficult, as shrubs vigorously resprout. We studied a combination of decapitation and honeysuckle leaf blight infection as a control effort. We measured plant responses at open and forested sites in northern and central KY, and regression was performed against the original basal diameter and day of the year (DOY). Shoot number, shoot length, total leaves, and total blighted leaves increased with diameter, while DOY consistently affected total blighted leaves. The percentage of blighted leaves also followed this trend across all sites. All sites saw a decrease in leaf density with DOY. Some of the smaller shrubs died, and logistic regression showed that shrubs with diameters 0.5 cm had 40% chance of death. Future work will focus on the long-term survival of decapitated shrubs.
Student Author(s): Kaitlyn Snyder, Trey Zinsmeister, Caitlyn Helton, Lauren Taylor
Faculty Author(s): Dr. Rick Boyce
Author Department(s): Biological Sciences
Title: Spatial ecology of Callery pear in northern Kentucky and southwest Ohio
Abstract: Callery pear (Pyrus calleryana) is a non-native species that has become invasive following its introduction to the United States as an ornamental tree. Previous work has suggested that Callery pear is mid-shade tolerant or shade-intolerant based on its stand structure. The shade tolerance of plants can also be determined by examining their spatial distribution. We selected several stands dominated by Callery pear in northern Kentucky and southwest Ohio. At each site we laid out a 10 m x 10 m grid and recorded the position of each Callery pear tree within the grid. In addition, we measured the diameters. We then determined Ripley's K to look for patterns of distribution and associations among size classes at each stand. In general, Callery pears were randomly distributed, with no association between size classes. This distribution follows the pattern typically seen in shade-intolerant plants with bird-distributed seeds.
Student Author(s): Serigne Seye*
Faculty Author(s): Warunya Panmanee, PhD, and Erin Strome, PhD
Author Department(s): Biological Sciences
Title: A Comparative Study of Reactive Oxygen Species Levels Between Wildtype S. cerevisiae and SAM 1 and SAM 2 Mutants
Abstract: Genome stability is crucial in maintaining the functions of all life, and instability in its integrity can be catastrophic for organisms. In Saccharomyces cerevisiae the genes SAM1 and SAM2 encode the S-Adenosylmethionine (AdoMet) synthatases which produce AdoMet the main methyl donor necessary for cell viability. Deletion of these genes has been shown to alter genome stability, but the mechanism is unknown. Reactive Oxygen Species (ROS) are known to cause damage to DNA and other macromolecules. AdoMet is generated in the methyl cycle which further leads into production of Glutathione, a molecule which neutralizes ROS the and can be protective to the genome. In this study we use an H2DCF-DA assay to measure the levels of ROS in wildtype S. cerevisiae compared to SAM1 and SAM2 mutants to determine how the expression of these genes can affect ROS levels which may provide insight on the role of these genes in genome integrity.
Student Author(s): Zachary Case
Faculty Author(s): Warunya Panmanee, PhD, Erin Strome, PhD
Author Department(s): Biological Sciences
Title: A Methodology for Measuring Loss of Heterozygosity rates in Saccharomyces cerevisiae
Abstract: S-Adenosylmethionine (SAM) is the second most commonly used cellular enzyme substrate. Differences in expression levels of the three human SAM synthetase genes, MAT1, MAT2a, and MAT2b have been found in several human cancers. Understanding the effects of these genes in the development and progression of cancer may reveal new treatment possibilities. Therefore, we adapted the methodology of Anderson et al. (2008) to study the effects of the paralogous SAM1 and SAM2 genes on the rates of loss of heterozygosity (LOH) in Saccharomyces cerevisiae when compared to wild type. LOH can contribute to the development of cancer.
Student Author(s): Amber Miller, Bridget Murray, Chris Ernst
Faculty Author(s): Dr. Allison Parker
Author Department(s): Biological Sciences
Title: Tick Species Distribution in the Northern Kentucky Region
Abstract: Ticks are one of many known vectors of diseases that can infect humans and other animals. There are six known tick species found in the state of Kentucky. However, the seasonality and distribution of these tick species is not well known in the Northern Kentucky region. The purpose of this study is to examine the distribution of tick species in the region over time. Over ten weeks, ticks were collected once a week at each of five rural sites. To date, 38 ticks have been collected. Thirty-seven collected ticks were American dog ticks, Dermacentor variabilis, and one collected tick was a Lone Star tick, Amblyomma americanum. Sampling will continue through August 2024. Understanding tick species distribution and seasonality in our region allows for effective control strategies that can reduce the occurrence of tick bites and tick-borne diseases in humans and other animals.
Student Authors: Sydney Bowman, Emily Collier, Abilene Morgan, Marisol Reyes
Faculty Author: Dr. Allison Parker
Author Department(s): Biological Sciences
Title: Seasonal Abundance of Mosquitoes Inhabiting Tires in Northern Kentucky
Abstract: Mosquitoes are vectors of diseases that infect humans and other animals and are also pests. Mosquito larvae develop in habitats with standing water, including human-made containers. Improperly discarded tires are an excellent habitat for mosquito larval development. The purpose of this study was to determine the mosquito species composition of mosquitoes found in discarded tires. Mosquito larvae and pupae were collected from tires at six sites in Boone County, KY weekly for ten weeks from June through August 2022. All collected mosquitoes were identified to species. Ten species of mosquitoes were found in the tires with over 4,200 individual mosquitoes collected and identified. The most commonly collected species was the Asian bush mosquito, Aedes japonicus, which accounted for over 60% of the collected mosquitoes. Understanding the density and distribution of mosquitoes in the region can help aid in targeted control strategies and campaigns to reduce improperly discarded tires.
Student Author(s): Sydney Bowman, Emily Collier, Abilene Morgan, Marisol Reyes
Faculty Author(s): Dr. Allison Parker
Author Department(s): Biological Sciences
Title: Unwanted Residents of Boone County: Mosquitoes
Abstract: Our lab research team spent the months of June to August 2022 studying the variety of mosquitoes present in areas of Boone county. Community awareness of species diversity is important because different mosquitoes can carry different dangerous diseases. We looked at mosquitoes at 6 different sites across the county by extracting larvae and pupae from tires that had water and debris accumulated inside them. We then identified them down to species level and kept track of the number of each species at each site. The most common species found were Culex restuans, Aedes japonicus, and Aedes triseriatus. These results are important for residents of the area to understand the organisms they are living near and the potential dangers they may pose. Mosquitoes are known to lay their eggs in containers of water and residents should be aware of this so they can avoid allowing mosquitoes to breed.
Student Author(s): Alex Walsh, Briannia Quarles, Kalyani Abbaraju, Mickayla Kowalski, Kevin Berling, Connor Perry, Angela Kyntchev, Susan Martin, India Davis, Karlee Migneault, Mackenzie Feltner
Faculty Author(s): Christine Perdan Curran
Author Department(s): Biological Sciences, Psychological Science, School of Computing and Analytics
Title: The Effect of Prenatal Benzo[a]pyrene Exposure on Motor Functions in CYP1A1 (-/-) Knockout and Wild Type Mice
Abstract: Benzo[a]pyrene (BaP) is a polycyclic aromatic hydrocarbon and known carcinogen which was recently found to be neurotoxic. Resulting from incomplete combustion reactions, it is inhaled in cigarette smoke and car exhaust, and ingested by consumption of grilled foods. CYP1A1 is an enzyme involved in the metabolism of BaP, to determine if genetic differences alter susceptibility to developmental BaP exposure, we treated pregnant Cyp1a1 (-/-) knockout and wild type mice with BaP during late gestation until pups were weaned. We examined the early development of reflexes and other motor functions in the mice by conducting surface righting reflex tests on postnatal days 5, 7 and 10 as well as negative geotaxis tests on postnatal days 7, 10, and 14. These mice later underwent rotarod tests as adults to assess their balance, coordination and motor learning.
Student Author(s): Kalyani Abbaraju, Alex Walsh, Briannia Quarles, Mickayla Kowalski, Kevin Berling, Connor Perry, Angela Kyntchev, Susan Martin, India Davis, Karlee Migneault, Mackenzie Feltner
Faculty Author(s): Christine Perdan Curran
Author Department(s): Biological Sciences
Title: Testing the Effect of Benzo[a]Pyrene Exposure During Early Brain Development on Memory in Mice
Abstract: Benzo[a]pyrene (BaP) is a polycyclic aromatic hydrocarbon found in traffic-related air pollution, cigarette smoke and grilled foods. Exposure to BaP can affect memory, motor skills and learning. We use a mouse model to determine if genetic differences change susceptibility to developmental BaP exposure and if early life interventions can mitigate the adverse effects. We used the Morris water maze to test hippocampal-dependent spatial learning and memory. An escape platform is placed in a pool of water and obscured with non-toxic white paint. Visual cues around the room are used to navigate to the platform. There are four weeks of testing. Each week, the platform is moved to a new location and a smaller platform is used, increasing the difficulty of the test. We measure the latency to find the escape platform, the distance traveled to the find the platform, and the average distance from the platform location.
Student Author(s): Briannia Quarles, Alex Walsh, Kalyani Abbaraju, Mickayla Kowalski, Kevin Berling, Connor Perry, Angela Kyntchev, Susan Martin, India Davis, Karlee Migneault, Mackenzie Feltner
Faculty Author(s): Christine Perdan Curran
Author Department(s): Biological Sciences, Psychological Science, School of Computing and Analytics
Title: Can exercise mitigate the adverse neurological effects of early life exposure to benzo[a]pyrene?
Abstract: Humans have widespread exposure to the polycyclic aromatic hydrocarbon benzo[a]pyrene (BaP) through vehicle exhaust, cigarette smoke, and grilled food. Exposure to BaP in the womb and during early life can lead to neurological deficits such as a lower IQ in humans. We used mice to model the effects of BaP on memory in wild type mice and those lacking the enzyme CYP1A1 which can metabolize BaP. Pregnant dams were treated with either the corn oil control or BaP dissolved in corn oil from gestational day 10 through weaning at postnatal day 25. We tested the memory of offspring at 2 months of age in Novel Object Recognition. To determine if exercise could improve learning and memory, we provided running wheels to dams, pups or both and compared results with control mice that didn't run.
Student Author(s): Sajida Ely Maouloud*, Lauren Wessel*, Seth Gregg, Paige Landfried
Faculty Author(s): Emily Shifley
Author Department(s): Biological Sciences
Title: The FGF signaling pathway plays an important role in Xenopus pharyngeal development
Abstract: The pharynx is a region in the vertebrate embryo that gives rise to craniofacial features, thymus and parathyroid glands and when it does not develop properly, birth defects can occur. The Fibroblast Growth Factor (FGF) pathway is one of the genetic signaling pathways that guide the development of the pharynx. We hypothesized that blocking the FGF pathway would disrupt pharyngeal development. We collected embryos, inhibited FGF signaling at different developmental stages, and analyzed them with western blotting, in situ hybridization, sectioning, and skeletal preparations. We found that certain genes like pax1 and cyp26a1 had reduced gene expression in the pharynx of FGF inhibited embryos, while other genes were unaffected. We also found that the craniofacial cartilage of FGF inhibited tadpoles was disorganized and reduced. This data shows that certain genes are downstream of the FGF pathway and that the actions of these genes are important for proper craniofacial development.
Student Author(s): Seth Gregg*, Paige Landfried*, Lauren Wessel, Sajida Ely Maouloud
Faculty Author(s): Emily Shifley
Author Department(s): Biological Sciences
Title: Iroquois genes act downstream of the FGF pathway in the developing pharynx
Abstract: The Fibroblast Growth Factor (FGF) genetic signaling pathway helps guide early embryonic development of the pharynx, which forms a number of important structures in the head and neck. The pharynx is composed of three germ layers, the endoderm, mesoderm, and ectoderm. We identified target genes of the FGF pathway called the Iroquois (Irx) transcription factors and hypothesized they play an important role in pharyngeal development. We manipulated Irx levels in developing Xenopus embryos by injecting morpholinos and examined gene expression with in-situ hybridization and sectioning. We looked for changes in expression levels of pax1, sdf1, and spry, which are found in the different layers of the developing pharynx. Our data will help identify which genes are downstream of Irx and FGF and how they work to coordinate pharyngeal development. These results, and future studies, can help identify genetic causes for pharyngeal birth defects and possible treatment or prevention methods.
Student Author(s): Sumaiya Mohamed*, Kailee Kenwright, Amila Coric
Faculty Author(s): Erin D. Strome, PhD
Author Department(s): Biological Sciences
Title: SAM2 Knockouts In Saccharomyces cerevisiae Impact AdoMet Production Which Alters Growth In The Presence Of Urea
Abstract: The S-Adenosyl-Methionine (AdoMet) synthetase genes, SAM1 and SAM2, show altered genome stability when mutated in S. cerevisiae cells. In a search to understand the causative link and in the hunt to find other cellular changes due to these mutations, these strains were subjected to more than 1400 conditions and measured for growth variation. Strains with no copies of SAM2 showed differences in growth in media with urea added. Cells use urea as a nitrogen source, and we seek to understand the cause of the altered growth in relationship to AdoMet production and within the context of RNA-sequencing data of differential gene expression in these cells.
Student Author(s): Lindsey Warinner*, Kailee Kenwright, Amila Coric
Faculty Author(s): Erin D. Strome, PhD
Author Department(s): Biological Sciences
Title: Exploration Of The Effects Of SAM1 And SAM2 Knockout Mutations In Saccharomyces cerevisiae In The Presence Of Hygromycin B
Abstract: SAM1 and SAM2 are both S-Adenosyl-Methionine (AdoMet) synthatases in Saccharomyces cerevisiae, but the deletion of these genes cause opposite impacts on chromosomal stability. In seeking to understand how these mutations impact chromosome stability we characterized a wide range of other phenotypes of these mutants. We observed that sam1-deficient cells showed increased sensitivity to Hygromycin B. It has previously been shown that glycosylation mutants are sensitive to aminoglycosides, therefore we sought to determine if our SAM1 mutation had impacts to glycosylation. We used gene ontology and genome databases to map out the genes of the glycosylation pathways, then overlaid RNA-Sequencing data to determine differentially expressed genes in the SAM1 mutants. We found there was no statistically significant differential expression of genes in the N-linked glycosylation pathway. This likely rules out changes in this glycosylation pathway being related to the observed growth differences, leaving 4 glycosylation pathways to be explored.
Student Author(s): Vy Ngo*, Kailee Kenwright, Amila Coric
Faculty Author(s): Erin D. Strome, PhD
Author Department(s): Biological Sciences
Title: Exploring the Impacts of Chloroquine on the Methyl Cycle in Saccharomyces cerevisiae Related to Iron Deficiency
Abstract: The absence of the SAM1 and SAM2 genes, which encoded for AdoMet synthetase enzymes in the methyl cycle, has been found to alter chromosome stability in Saccharomyces cerevisiae. In this study, these mutants were introduced to Chloroquine, an antimalarial medicine and a potential anti-cancer treatment, and measured for growth variation. Chloroquine is known to cause iron deficiency in S. cerevisiae and caused different impacts on the growth patterns of SAM1 and SAM2 mutant strains when compared to wildtype cells. We seek to understand the cause of this altered growth, relative to AdoMet changes, and differential gene expression in our mutant strains.
Student Author(s): Chrisula M. Stone, Rebecka L. Brasso
Faculty Author(s): Lindsey A. Walters
Author Department(s): Biological Sciences
Title: A Bander's Contribution to Ecotoxicology: Comparison of Mercury Concentration in Feathers and Blood of Nestling Tree Swallows Demonstrates Feathers' Suitability as a Biomonitoring Tool
Abstract: More tools are needed for ecotoxicologists to assess patterns of organismal mercury uptake. The tree swallow (Tachycineta bicolor) is a bird that readily takes to nest boxes and is known to bioaccumulate mercury in its tissues. We sought to define the relationship between blood and feather mercury in nestlings. We collected blood and feather samples during banding for analysis. We found a significant positive relationship between blood and feather mercury, and that significantly more mercury was found in the feathers than blood, demonstrating the potential of feathers collected during banding as a biomonitoring tool.
Student Author(s): Tami Farber
Faculty Author(s): Lindsey Walters
Author Department(s): Biological Sciences, School of Computing and Analytics
Title: Climate and the reproductive success of Tree Swallows and Eastern Bluebirds
Abstract: Temperature regulation during incubation is a key factor in the breeding success of most birds. A nesting parent spends energy to maintain the necessary egg temperatures, at times to their own detriment. Our changing climate has led to changes in surface and air temperatures, which could affect the energy cost for birds of incubating eggs. To test whether differences in temperature affect the number of successful offspring for the season, we compared five years of local temperature data to five years of data from nest box monitoring at two locations near Cincinnati, OH. We did not find a significant relationship between temperatures during incubation and the total number of Tree Swallow and Eastern Bluebird nestlings that successfully fledged, which could mean that the birds are able to adapt their reproductive behavior to changing temperatures. However, a different approach to the data in future studies could reveal more.
Student Author(s): Ashley Denney
Faculty Author(s): Lindsey Walters
Author Department(s): Biological Sciences
Title: Who does more diaper duty? Comparison of fecal sac removal rates of male and female Tree Swallows.
Abstract: The removal of fecal sacs, a mucus covered waste product of nestlings, is a subject that is relatively under-investigated for Tree Swallows (Tachycineta bicolor). Both parents take turns removing the sacs, disposing of them far from the nest. We predicted that male Tree Swallows would remove more fecal sacs than females because the female spends time brooding on the nest. We recorded the number of times each sex removed a fecal sac during 1 hour and found that there was no significant difference between the overall removal rate for males and females. However, there was a significant interaction between sex and nestling age on fecal sac removal rate, with males removing more when nestlings were young and females removing more after day 8. These results can be used to gain insight into the behavior of Tree Swallows as well as create predictions for other bird species that have similar behaviors.
Student Author(s): Presley Riggs
Faculty Author(s): Lindsey Walters
Author Department(s): Biological Sciences
Title: Time of Day has No Impact on Tree Swallow Nestling Provisioning Rates
Abstract: This experiment focused on the relationship between the nestling provisioning of Tree Swallows (Tachycineta bicolor) and the time of day. We predicted that Tree Swallows would provision more frequently earlier in the morning due to flying insects being more active then. We acquired the necessary data by doing hour-long observation periods of Tree Swallow nest boxes at Middleton-Mills Park, recording data on visitation rates for each nest box. The results revealed that there is no significant relationship between time of day and Tree Swallow provisioning rates, disproving our hypothesis. Future research on these variables should be more in-depth with more nest boxes to minimize variation of individual parents.
Student Author(s): Megan Bunnenberg
Faculty Author(s): Lindsey Walters
Author Department(s): Biological Sciences
Title: Evidence of conditional cooperation in Eastern Bluebird parental care
Abstract: Animal biparental care creates conflict, since it is advantageous for both parents to do the least amount of work for the greatest amount of fitness. A basic form of reciprocity, where parents match each other's feeding rate, may help equalize parental investment or increase overall provisioning rate for offspring. Our experiment sought to determine if this is occurring in Eastern Bluebirds. We experimentally increased female provisioning rate by using a Bluetooth speaker to increase the number of nestling begging calls she heard. We then observed whether the male provisioning rate changed in response. We found that males significantly increased their provisioning when females did. This indicates that Eastern Bluebird parents are adjusting their own behavior based on their partners' parental effort. Reciprocity is rare in animals, and finding evidence of this in Eastern Bluebirds may illuminate another factor that may be involved in the evolution and shaping of biparental care.
Student Author(s): Bree Sweeney, Jenna Dunham
Faculty Author(s): Lauren Williamson, PhD
Author Department(s): Biological Sciences
Title: Effects of Sex and Inflammation on Context Object Discrimination Memory in Rats
Abstract: Inflammation can have profound effects on the brain and memory. We assessed the effects of sex and inflammation on rat learning and memory with the context-object discrimination (COD) task. The COD task requires rats to recognize a familiar object in an unfamiliar setting after two sessions of exploration. Inflammation affects males and females differently, with an increase in inflammatory signaling in males that is not present in females. The increased inflammation disrupts male memory formation on the COD task. We tested the rats' memory on the COD task after treatment with either LPS or saline. Serum levels of estradiol and testosterone were then analyzed to evaluate possibly hormonal mechanisms for the sex differences in behavior.
Student Author(s): Kyle Winkler, Ben Coburn
Faculty Author(s): Lauren Williamson, PhD
Author Department(s): Biological Sciences, School of Computing and Analytics
Title: The Effects of Helminth Treatment and Inflammation on Hippocampal Dependent Fear Conditioning Memory
Abstract: Chronic inflammatory diseases have been on the rise and commensalist parasites may provide anti-inflammatory signals that attenuate disease-causing inflammation. Early life inflammation has a significant effect on behavior, especially when combined with inflammation in adulthood (Williamson et al., 2016). This study assesses the interactions between helminth (commensalist parasite) treatment, early-life and adult inflammation, and sex on hippocampal dependent memory using context pre-exposure facilitation effect (CPFE) fear conditioning. Pups then received a 3-stage treatment with PBS or E. coli on postnatal day 4, then at weaning offspring were treated orally with H. diminuta or saline. The final treatment was administered during day 1 of CPFE through an injection of saline or lipopolysaccharide causing a double-hit of neuroinflammation. Thus far, we've only seen a sex-dependent difference in scored behaviors. This study will allow us to further understand immune stressors and possible interventions, which later could help aid medicine and human health.
Student Author(s): Madeline Buroker, Erin Matthews
Faculty Author(s): Lauren Williamson, PhD
Author Department(s): Biological Sciences
Title: Affects of Maternal Helminth Inoculation on Rat Pup Neuroinflamation
Abstract: Immune challenges in early life can greatly impact neurodevelopment. To determine if the pre- and postnatal environments may affect early life development, female rats were treated orally with either rat tapeworms (H. diminuta) or saline before conception. Brain tissue was collected from both sexes in each litter on postnatal day one (P1) and P4. The remaining pups received injections of either E. coli or PBS on P4, and brains were collected from one male and one female on P7. This tissue was sliced and stained with Iba1. Using stereology, microglia, the primary immune cells in the brain, were counted in the hippocampus and classified by morphology (e.g., round, stout, thick long, and thin long). In P1 and P4, pups born to dams treated with helminths had delayed microglia maturity compared to pups born to dams treated with saline. We predict that these differences will not be present in P7 brains.
Student Author(s): Men Tran*
Faculty Author(s): Warunya Panmanee, PhD, Erin Strome, PhD
Author Department(s): Biological Sciences
Title: Investigating Intracellular ATP Levels In The Yeast Cell Saccharomyces cerevisiae Upon The Knockout Of AdoMet Synthetase Genes, SAM1 And SAM2
Abstract: S-Adenosyl-Methionine, AdoMet, plays a significant part in the methyl cycle in all organisms as the main methyl group donor. The genome of budding yeast, Saccharomyces cerevisiae, has been found to experience changes in stability when their AdoMet synthetase enzymes, SAM1 and SAM2, are knocked out. Further, changes in AdoMet levels were documented. ATP and methionine are the main components of the AdoMet synthesis process. Therefore, this project aims to explore how intracellular ATP concentration might alter due to these gene mutations. This measurement will give us a better understanding of the effect of changes in AdoMet production on other cellular processes in budding yeast.
Student Author(s): Elizabeth Durham, Olivia Onodu, Njideka Nnorom, Fiona Chen, Rafael Verduzco
Faculty Author(s): Charlisa Daniels
Author Department(s): Chemistry and Biochemistry, Biological Sciences
Title: IDA-Functionalized Ion-Selective Films Aim to Harvest Valuable Ions for Sustainable Industrial Wastewater Filtration Applications
Abstract: The goal of this investigation was to develop a membrane-based process selective toward Cu by fabricating polymeric membranes with ion-specific groups, quantifying the permeability and selectivity towards Cu ions, implementing membranes in a continuous-flow process for Cu removal, and maximizing energy efficiency and minimizing the cost of Cu removal and recovery. Primary components under investigation for these monomers were the GMA, IDA, PEGDA, and PEGMEA. Assumptions were that the 50% and 60% water contents for the films were best; however, 50% water content films produced unacceptable results, revealing air bubbles and phase separation. To address this, solely 60% water content was used and water weight was fixed for the 5 and 10 mole percent samples. Trials were run using water sorption tests, gel fractions, and conductivity calibrations. The membranes were exposed to NiCl2, MgCl2, and CuCl2. After exposure, samples of the ion solutions were taken and analyzed via ICP Spectroscopy.
Student Author(s): Ashton J. Davey
Faculty Author(s): Michael P. Guy
Author Department(s): Chemistry and Biochemistry, School of Computing and Analytics
Title: A bioinformatics approach for the identification of a box C/D guide RNA responsible for the Nm39 modification of tRNA
Abstract: tRNA modifications play an important role in translation. The enzyme responsible for the 2-O methylation of tRNA position 39 is unknown and only occurs in multicellular eukaryotes. Prior studies in our lab have tested orphan methyltransferases but have not yet found the responsible enzyme. Alternatively, this modification could be carried out by a box C/D guide RNA. I used PLEXY, a computer tool, to identify guide RNA targets. I tested three tRNAs known to receive the Nm39 modification in humans. Candidate guide RNAs included known orphan Box C/D guide RNAs. With the current dataset, no results have corresponded to a modification of position 39. This indicates that the tested guide RNAs are not responsible for the modification or may not function in the same manner as the majority of guide RNAs. Further research may use different tools and an expanded list of candidate guide RNAs and tRNAs.
Student Author(s): Samuel Seibert, Linh Le
Faculty Author(s): Michael P. Guy, Holly Funk
Author Department(s): Chemistry and Biochemistry
Title: Cloning and expressing Trm7, Trm732, and Trm734 proteins to examine binding interactions with tRNA-Phe
Abstract: Posttranscriptional modifications in the form of 2`-O-ribose methylations occur at the anticodon loop of tRNA-Phe in Saccharomyces cerevisiae (brewers yeast). The methyltransferase protein Trm7 achieves methylation at nucleotides C32 and G34 by associating with Trm732 and Trm734, respectively. Defects in Trm7 causes a sick phenotype in brewer's yeast. Likewise, defects in the Trm7 human ortholog, FTSJI, causes intellectual disability. Trm7 plays a catalytic role in methylation activity, and Trm732 and Trm734 are predicted to bind and position tRNA-Phe for methylation. We are cloning protein expression plasmids containing the TRM7, TRM732, or TRM734 genes. Plasmid expression in Escherichia coli will allow us to purify our target proteins. Through in vitro protein-tRNA binding assays, we will define whether Trm732 and Trm734 can bind to tRNA-Phe. Previously identified Trm732 and Trm734 variants mutated at functionally important amino acid residues will be analyzed in a similar manner to define their affinity for tRNA-Phe.
Student Author(s): Alexandra K. Mullins, Natalie N. Creech, Mikayla Case
Faculty Author(s): Michael P. Guy, Holly M. Funk
Author Department(s): Chemistry and Biochemistry, Biological Sciences
Title: Strategies for identifying important residues in the tRNA modification protein Trm732
Abstract: Post-transcriptional tRNA modifications are required for efficient protein translation. In yeast, the Trm7 methyltransferase forms a complex with Trm732 to modify tRNA at position 32. In humans, lack of Trm7 causes intellectual disability. Little is known about the function of Trm732, although we have shown that one conserved motif in Trm732 is important for tRNA modification. We are identifying other residues important for function using two strategies. First, we compare Trm732 proteins of different organisms using protein alignments to determine conserved regions to mutate for testing. We have generated four new Trm732 variants and are testing their function. Second, we are expressing randomly mutated Trm732 variants in a sick strain that lacks Trm732. Lack of rescue of the strain indicates that the mutation present is harmful to protein function. DNA from these colonies will be extracted and sequenced to determine the mutation that causes loss of Trm732 function.
Student Author(s): Nick Korzenborn
Faculty Author(s): Michael P. Guy, Holly Funk
Author Department(s): Chemistry and Biochemistry
Title: Identification of key residues in the function of Trm734 via random and site-directed mutagenesis
Abstract: In Saccharomyces cerevisiae, 2'-O-methylation of nucleotides C32 and G34 in the anticodon loop of tRNAPhe is necessary to facilitate translation. Trm7 complexes with accessory protein Trm734 to perform this post-transcriptional modification of G34, but little is still known about the function of Trm734. Recently, we identified three key motifs required for Trm734 function. Two methods are being used to further define key residues and motifs in Trm734: random mutagenic PCR and site-directed mutagenesis. A new protein alignment of 9 diverse eukaryotic species was used to identify conserved residues and motifs within Trm734 as targets for site-directed mutagenesis. This alignment identified motifs TDG835, QSG882, and GDD905 as important for Trm734 function. These motifs will be replaced with AAA via site-directed mutagenesis and tested in a strain lacking functional Trm734 to assess importance for Trm734 function. We are also currently working to generate the yeast strains required for the random mutagenesis experiments.
Student Author(s): Thao Linh Le
Faculty Author(s): Michael P. Guy
Author Department(s): Chemistry and Biochemistry
Title: Determination of the effect on the translation of 2'-O-methylation at positions 32 and 34 in eukaryotic tRNA
Abstract: A cause of non-syndromic X-linked intellectual disability is mutated FTSJ1, a methyltransferase responsible for post-transcriptional 2'-O-methylation of tRNAPhe. The yeast ortholog of FTSJ1 is Trm7, which interacts with Trm732 and Trm734 to perform 2'-O-methylation at position Cm32 and Gm34 on tRNAPhe, respectively. Due to wobble pairing with position 34, tRNAPhe can read both UUU and UUC codons. In trm7 mutant yeast, lack of 2'-O-methylation at G34 and C32 causes translational deficiency. However, the individual roles of Cm32 and Gm34 in tRNAPhe function are unclear. The RNA-ID reporter system can help study dependence of UUU and UUC codons on Cm32 and Gm34, by allowing bidirectional expression of red fluorescence proteins (RFP) and green fluorescence proteins (GFP). Codons to be tested are inserted in front of GFP, allowing translation analysis in cells by their RFP/GFP ratio. We are currently working to integrate the reporter into trm732 and trm732 mutants prior to experiments.
Student Author(s): August Bozarth, Gabrielle Kerr, Lillian Lown
Faculty Author(s): Patrick M. Hare
Author Department(s): Chemistry and Biochemistry, Biological Sciences
Title: Further Investigation of the Photo-degradation of Estrone and the Synthesis of Lumiestrone
Abstract: Increased amounts of estrone in water causes detrimental effects to wildlife, especially fish. When estrone is photolyzed, it generates a product called lumiestrone with the rate of production being much slower in water than in acetonitrile. To understand the cause of this, the photolysis of estrone was carried out in varying percent volumes of water. While this product has been identified in nature from photolysis, to more easily study it, it was synthesized and purification methods were used to yield consistent results. The properties of lumiestrone were tested by UV-Vis spectroscopy, fluorescence spectroscopy, and proton NMR. An attempt at quantitative analysis of the photoproducts of lumiestrone was performed by both gas chromatography-mass spectrometry and high resolution liquid chromatography-mass spectrometry. However, differences in product properties and some synthetic byproducts interfered.
Student Author(s): Elma Coric, August Bozarth
Faculty Author(s): Patrick M. Hare
Author Department(s): Chemistry and Biochemistry
Title: Measurement of Estrogens and their Photodegradation using Raman and SERS Spectroscopy
Abstract: Estrogenic pollutants can potentially be detected through Raman spectroscopy and treated through photolysis. Raman spectra of solid estrogenic compounds (estrone, 17²-estradiol, lumiestrone, and estrone-3-methyl ether) were collected to distinguish the estrogens and their photoproducts. Dissolved samples were also tested, but micromolar concentrations of estrone were difficult to detect through Raman spectroscopy. To enhance signals in solution, SERS was utilized to measure the photo degradation of estrone in acetonitrile. Both a silver nanoparticle SERS solution and solid gold nanoparticle SERS plates were compared, with only the silver giving results. More work on increasing signal intensity is needed before SERS can be used to quantify photodegradation in solution.
Student Author(s): Kelsey Bose, Eric Jump, Thu Nguyen, Kendra Baker
Faculty Author(s:) Lili Ma
Author Department(s): Chemistry and Biochemistry, Psychological Science
Title: ¿ªÐÄ¹í´«Ã½ UR STEM: Domino Synthesis of Heteroaryl Compounds as Potential Breast Cancer Treatment
Abstract: Breast cancer is one of the most common diseases among women worldwide. The current synthetic methods to prepare anti-breast cancer drug molecules such as Lextrazole, require several steps and creates a lot of chemical waste. This study is focused on the development of a new class of small molecules as anti-breast cancer drug candidates via a highly efficient domino reaction. The reactive properties of the palladium catalyst, XPhos G4 Pd was used to expedite the reactions between ketones (2-methyl-1-phenylpropane-1-one I16, 3-acetyl-2,5-dimethylthiophene I52, 2-acetylpyridine I53, 4-acetylpyridine I54, acetophenone C0) and heteroaryl halides (methyl 3-iodopyridine-4-carboxylate HIH401, methyl 3-bromopyridine-4-carboxylate H401). The reaction mixtures were exposed to microwave irradiation at 130oC, 160oC, or 180oC for either 20 or 30 minutes. By using the domino method, the reaction time was greatly shortened, and chemical waste was minimized. The developed synthetic route in this study shows great promise in making chemical reactions more efficient.
Student Author(s): Kristian Kowsky, Hai Hoang, Lillyann Tursany
Faculty Author(s): KC Russell, Nicholas Caporusso
Author Department(s): Chemistry and Biochemistry, School of Computing and Analytics
Title: Enhancing the Chemistry Gaming Experience
Abstract: Our project explores the use of games as learning tools. Specifically, we focused on improving a card-matching studying game developed for teaching chemistry to university students. Our work involved redesigning the game to update its visual assets and to enable reusing its logic in other disciplines. To this end, we designed a step-by-step process for creating art assets. Specifically, we used modeling programs like JSmol to create the 3D models of chemicals and atomic orbitals, then captured the output as a video to convert the 3D model into a static asset for use in the studying game; finally, we produced static assets for twenty-two chemicals and ten orbitals. Simultaneously, we completed the back-end component of the game, which enables instructors to easily use the card-matching game as a template for other topics and disciplines by just designing and uploading new sets of assets. Finally, we identified additional game templates that could be incorporated into a repository that teachers can leverage in as learning material and quizzes for their courses.
Student Author(s): Olivia Villaflor, Khuyen Ho, Adrienne Hafley
Faculty Author(s): KC Russell
Author Department(s): Chemistry and Biochemistry
Title: Synthesis of Amino Acid-oxacalixarene Hybrids
Abstract: The goal of the Russell Research Group is to synthesize m4-oxacalixarenes with various functional side chains. My project focuses on the preparation of an amino acid-oxacalixarene hybrid that possesses an L-dihydroxyphenylanaline (L-DOPA) sidechain. A calixarene is a compound that is a compound in which benzene rings are bridged by a methylene group. When a methylene group is replaced with an oxygen, the compound is renamed as an oxacalixarene. To synthesize this compound, we have prepared 3,5-dihydroxyphenylaniline to make asymmetric annulenes with meta-linked oxacalixarenes. Amino acid-oxacalixarene hybrids have many possible applications including biological pathway inhibition involved in disease and molecular recognition. This poster will review the progress that has been made towards the synthesis of an amino acid-oxacalixarene hybrid.
Student Author(s): Evan R. H. Page, Miranda S. Shephard, Katherine E. Wilcher
Partner Author(s): Lisa M. Privette Vinnedge, PhD
Author Department(s)/Partner(s): Biological Sciences, Cincinnati Children's Hospital Medical Center - Department of Oncology
Title: DEK Deletion in DNA-pk Mutant Mice Exhibits Hyperproliferation and Hematopoietic Dysfunction
Abstract: Blood cells are created through a process called hematopoiesis. Mutation of the DNA-pk gene and aberrant expression of the DEK protein independently have been linked to defects in hematopoiesis. Cell culture studies indicated that dual loss of DEK and DNA-pk had negative consequences for cellular health. A DNA-pk mutant and DEK deficient mouse model was used to determine if the loss of both genes resulted in hematopoietic dysfunction or malignancy. We found that a lack of DEK expression within a DNA-pk mutant mouse model resulted in a significant increase in proliferation levels as detected by immunohistochemical staining for Ki67 in the bone marrow. This correlated with an increased white blood cell count. This novel comparison allowed for the quantification and visualization of hematopoietic functionality in the double-mutant mice and will serve as a tool to further investigate the role of DEK and DNA-pk in normal and malignant hematopoiesis.
Student Author(s): Neil Findley
Author Department(s): Mathematics and Statistics
Title: A. A. Albert's On a class of trinomial equations
Abstract: In the 1950s and 1960s, University of Chicago Algebraist A. A. Albert (1905 - 1972) participated in various mathematical research projects for the National Security Agency. Recently the reports that Albert wrote about those projects have been declassified. This project explores the results reported by Albert of his 1961 effort to find solutions in a field of degree n over the field of two elements for the equation 1 + x = xáµ. Several examples of solutions based on Albert's work are created.
Student Author(s): Roland Long
Author Department(s): Mathematics and Statistics
Title: Albert's Inversion Problem
Abstract: In the 1950s and 1960s, University of Chicago algebraist Abraham Adrian Albert (1905-1972) participated in various mathematical research projects for the National Security Agency. Recently the reports that Albert wrote regarding those projects have been declassified. This project explores the results of his work on the "inversion problem." The problem requires the determination of an n-tuple of 0s and 1s given certain relationships between the n-tuple and cyclic permutations of its components. Albert submitted five reports on this investigation between 1958 and 1964. Three reports develop combinatorial methods, using distances between the original n-tuple and its cyclic permutations to generate a suitable n-tuple. Two reports consider relationships between a polynomial representation of the n-tuple and a polynomial, which Albert called the norm of the n-tuple, to develop a computational method of generating the n-tuple.
Student Author(s): Andrew Code, Khue Dinh, Katrina Harp, An Ngo, Ethan Schnee, and Riya Shrestha
Author Department(s): Mathematics and Statistics, School of Computing and Analytics
Title: Predicting the Severity of California Wildfires
Abstract: The recent dramatic increase in wildfire severity in California has cost the insurance industry billions of dollars. To meet their regulatory obligation to remain solvent, insurance companies rely on predictions. However, wildfires are a particularly challenging peril to model because of the recent increases in severity and associated losses. We partnered with the American Modern Insurance Group to explore possible predictors of wildfire severity in California using publicly available wildfire and meteorological data. We created multiple linear regression models to predict wildfire severity, as measured by the number of acres burned, and then improved our results by introducing a logarithmic transformation. Lastly, we built classification models using machine learning algorithms that were able to classify summer wildfires as either small or large with decent accuracy.
Student Author(s): Wen Liu, Lauren Williams
Faculty Author(s): Gang Sun
Author Department(s): Physics, Geology, and Engineering Technology
Title: A PLC real-life application simulation LAB developing
Abstract: The use of Programmable Logic Controllers (PLCs) in industrial have gradually growth by each year for the past twenty years. Higher education institutions have paid more attention to developing their PLC class for EGT or MET program since 2000. Bring the real industrial environment and equipment's to the course and lab achieve hands-on experiences is one of the main philosophies for engineer education. However, due to the high cost of the real industrial PLCs and limit by the funding. Many institutions have come up with many different solutions. Some institutions use the computer software to simulate, some institutions decide to purchase low budget equipment and try to achieve a similar result, some institutions purchased few sets of equipment and divided students into groups to overcome this challenge. We are going to use the "easyPLC" software from Nirtec to develop two simulation LAB for the PLC course. There are some common industrial applications, conveyor and mixing tank are the two most common application around Northern Kentucky area. LAB one was selected to simulate conveyor system and LAB two was selected to simulate the mixing tank application.
Student Author(s): Leslie Ferrao
Faculty Author(s): Dr. Mahdi Yazdanpour
Author Department(s): Physics, Geology and Engineering Technology
Title: Designing a Brain-Computer Interface to Control a Mini Industrial Robotic Arm
Abstract: Integrating the electrical activities of the human brain with mechatronic systems to develop mind-controlled systems has become one of the most cutting-edge research topics in the Human-Robot Interaction and Robotics communities. In this research project, we designed and implemented an electroencephalography (EEG)-based brain-computer interface (BCI) to monitor and analyze human brainwaves. We use an EEG brainwear to capture the human brain wave patterns and send raw neuro signals to our proposed interface. This BCI recognizes changes in brainwaves when the user imagines performing a specific function. Our system processes and classifies the signals and converts them to meaningful commands used to control a mini-industrial robotic arm. This desktop robotic arm consists of 23 pieces and is completely printed using a fused filament fabrication (FFF) 3D printer. The whole mechanism is based on three stepper motors that allow the robot to produce the sequence of automatized movements on five axes. Our brain-computer interface enables users to control this mini-industrial robotic arm in different translational directions using the power of their minds.
Student Author(s): Manogya Aryal, Abby Jones, Justin Selby
Faculty Author(s): Seth Adjei
Author Department(s): School of Computing and Analytics
Title: Creating Instructional Elementary Python Programming Videos for Use in PLACEments
Abstract: PLACEments was created as an adaptive testing and remediation tool for students in mathematics. Students are given questions of varying difficulty based on their earlier answers and underlying skill prerequisites. This research extends PLACEments by including Elementary Python Programming content and is intended to apply to an introductory programming class at ¿ªÐÄ¹í´«Ã½. Each of the three researchers created instructional videos for forty different programming skills. The content of these videos ranged from running a basic program in Python to understanding and creating functions. Students would be asked elementary programming questions through PLACEments and a video will be shown to students who answered a question incorrectly. The researchers have peer-reviewed the videos to assess their quality. The effectiveness of this research is yet to be concluded as the videos need to be tested with students of Elementary Python Programming by measuring their improvement after viewing a video.
Student Author(s): Trang H. Do
Faculty Author(s): Dr. Alina Campan, Dr. Traian M. Truta
Author Department(s): School of Computing and Analytics
Title: A Time-Series Analysis of the Relation between Social Media and Stocks
Abstract: The massive data volume generated every day and growing computing power have enabled data-driven applications in almost every field. In this study, I applied time-series correlation measures to see if there is a relation between sentiment of Twitter posts, Tweet volume, stocks volume and stocks prices for 5 popular stocks from the S&P 500 index, with data collected from Twitter API and website. Results suggest that stocks prices are positively correlated with tweet sentiment, while negatively correlated with stocks volume and Tweet volume. I also compared the accuracy of different supervised and unsupervised machine learning techniques applied for tweet sentiment analysis, including Support Vector Machine, Valence Aware Dictionary and Sentiment Reasoner (VADER), Logistic Regression, transformer and textblob. Results show that VADER produces results best detect human sentiment with an accuracy of 52.74%. The research implies the usage of sentiment analysis in predicting stocks prices, and further work can be done to extend the analysis on 11 different sectors of the S&P 500 index.
Student Author(s): Alex Aossey, Trung Cao, Abhishek Shrestha, Brett Thaman
Faculty Author(s): Junxiu Zhou, Yangyang Tao, Nicholas Caporusso
Author Department(s): School of Computing and Analytics
Title: Gaze Tracking Technologies and Usage in Modern Research
Abstract: Eye-tracking technology has a history in medical science dating back even beyond computers, and as modern developments continue to refine human-computer interaction, researchers from a variety of fields have begun to explore potential uses of eye-tracking. Gaze is as much a measure of focus and attention as sight; with this in mind, gaze-tracking studies have been performed in the fields of education, psychology, design, sports science, accessibility, and marketing. In this work, we seek to analyze the current state of gaze-tracking technology, both by surveying experienced researchers and conducting interviews. With this gathered research, we aim to create a better understanding of the capabilities and limits of modern gaze-tracking technology, as well as how those boundaries might be pushed with future gaze-tracking solutions.
Student Author(s): Nahom Beyene
Faculty Author(s): Ankur Chattopadhyay
Author Department(s): School of Computing and Analytics
Title: Psybersecurity
Abstract: Psybersecurity is a term used to describe the threat posed by cyber-attacks on the human mind and the computer systems that control them. Dr. Louie stated that psychiatric engineering: is an attack vector that relies heavily on an individual's health. A potential way to protect OHI users from rising psybersecurity threats is by leveraging specialized online tools and technologies that can provide information assurance. Building a robust human-security network would safeguard people's psychological well-being more. With the rise of technology in every facet of daily life, the increased conveniences come with multiple security risks. The psybersecurity of OHI users is linked to their mental perceptions and state of mind. Building a strong community of individuals that check up on each other's mental states would help identify a behavior change. They suggest that within the domain of human security, psybersecurity is an emerging topic. It has plenty of scope for future work, as there has not been much research in this area.
Student Author(s): Natalie Brown
Faculty Author(s): Wei Hao
Author Department(s): School of Computing and Analytics
Title: Edge Computing and Cloud Computing Research
Abstract: Cloud computing is increasingly becoming a method of choice in computing solutions. With emerging technologies, new unforeseen issues arise with it. The computational performance and efficiency of cloud based apps are a particular aspect of interest. Analyzing and finding solutions to the performance of a cloud based can have huge impact in the time, computational complexity, and ease of use within cloud based systems. As a method of study for the research, we implemented comparative edge computing methods on the Amazon Web Services Cloud, (AWS). The performance of these methods were measured and analyzed to best approximate the abilities and advantages of differing edge computing methods. From these research study results we can find the best and most efficient solutions in solving for the performance and efficacy of cloud based apps.
Student Author(s): Zachary Daher, Brendan Martin, Connor Morgan
Faculty Author(s): Awad Mussa
Author Department(s): School of Computing and Analytics
Title: Mapping Mapping Zero-Click Attacks Behavior into MITRE ATT&CK Mobile Using CISA ATT&CK Mapping Guide
Abstract: A Zero-click attack is a sophisticated class of attacks where a smartphone device can be compromised remotely without any user interaction. Recently, there has been traces of successful zero-click attacks on iPhones by Pegasus, hacking spyware that can infect billions of phones running either iOS or Android operating systems. However, there is a lack of behavioral profile for a Zero-click attack, which is critical for a response and prevention plan and security awareness. To that end, we aim to explore leveraging CISA-MITER-ATT&CK-Mapping-Guide to map Zero-Click Attacks threat intelligence into MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) Mobile Matrix. Based on the forensic data from Amnesty's lab, we have discovered a total of 65 techniques used by the zero click attacks over the course of 13 different tactics. We have also compared CISA-MITER-ATT&CK-Mapping-Guide results with the expert's mapping and we have found many similarities between the two mappings.
Student Author(s): Abel Tsegaye, Khanh Nguyen
Faculty Author(s): Yangyang Tao
Author Department(s): School of Computing and Analytics
Title: Applied Machine Learning in Video Games
Abstract: Reinforcement learning in video games has become an interesting and promising direction on how to understand and implement machine learning in general. There are different algorithms, but the three algorithms: Neuro Evolution of Augmenting Topologies (NEAT), Q-learning and Policy gradient will be implemented on the video game Flappy Birds. After the algorithms are implemented, the computer will learn how to play the video game by itself after several trials and errors. Neuro Evolution of Augmenting Topologies (NEAT) is the first algorithm integrated as a prototype. It has two parts, the game body, and the machine learning algorithm. Later, we implemented Q-learning and Policy Gradient and tested the performance of the three algorithms to check on their efficiencies. The main purpose of this project is to implement those three algorithms and also to understand which algorithm yields a better performance.
Student Author(s): Murtadha Almakki, Nga Phan
Faculty Author(s): Dr. Yangyang Tao, Dr. Junxiu Zhou
Author Department(s): School of Computing and Analytics
Title: Face Recognition with Unlabeled Dataset
Abstract: Face recognition systems can normally attain good accuracy when they are provided with a large set of training examples. However, when a large training set is not available, their performance is commonly poor. In this work we explore methods for face recognition that achieves good performance when only a very small training set is labeled. The key idea is augmenting the original training set with previously unlabeled data. Specifically, we apply the well-known eigenfaces technique to reduce the dimensionality of the image space, then we perform an iterative process, classifying all the unlabeled data with an ensemble of classifiers built from the current training set, and appending to the training set the previously unlabeled examples that are believed to be correctly classified with a high accuracy, according to the ensemble. We experimented with ensembles based on the k-nearest-neighbors, feedforward artificial neural networks and locally weighted linear regression learning algorithms.
Student Author(s): Khanh Nguyen, Abel Tsegaye
Faculty Author(s): Yangyang Tao
Author Department(s): School of Computing and Analytics
Title: Investigate Application of Machine Learning Algorithms in Video Games
Abstract: The application of reinforcement learning in video games has become one of the most interesting and promising directions to understand and interpret the reinforcement learning. However, there are fewer studies on the comparison between different algorithms. In this project, three reinforcement learning algorithms: Neuro Evolution of Augmenting Topologies (NEAT), Q-learning, and Policy Gradient will be integrated into video games such as Flippy Birds. Then we will teach the computer to learn how to play the video games by themselves. Specifically, we first implement the NEAT into the game as the prototype. The prototype has two parts. One is the game body; the other is the learning algorithm. Then we implemented the Q-learning and Policy Gradient into the game based on the prototype and tested the performance of the three algorithms. The main purpose of this project is to implement these three algorithms.
Student Author(s): Wayne Leeke
Author Department(s): School of Computing and Analytics
Title: Manipulating Classification Models with GANs
Abstract: Previous research has shown that artificial neural networks can be attacked by generating inputs that manipulate the saliency of a model. Saliency being the importance of individual features in making specific categorical decisions. These adversarial inputs minimize or maximize a gradient of these features to generate an input that sits right outside or in a decision boundary. Crafted inputs ideally are indistinguishable to both the model and the human eye. Gradient attacks can be mitigated by filtering or retraining the model on the adversarial inputs flagged as malicious. Alternative attacks based on generative adversarial networks seem effective against simple feed forward network architectures. Malicious GAN models can be retrained against the targeted model to defeat common adversarial input mitigations. More research is needed to determine if this technique is affective for more complex architectures such as recurrent and convolutional networks.
Student Author(s): Nga Phan, Murtadha Almakki
Faculty Author(s): Dr. Junxiu Zhou, Dr. Yangyang Tao
Author Department(s): School of Computing and Analytics
Title: Machine Learning for Face Recognition in Small Dataset
Abstract: Face recognition attracts a lot of research attention due to its wide applications such as surveillance, identification, forensics, and human-computer interactions. With the increasing availability of machine learning algorithms such as deep learning, most existing research uses massive datasets to facilitate better recognition performance. However, it is time-consuming and labor intensity to collect a large face dataset. With a small dataset, some of the machine learning algorithms may fail to perform the recognition task. This work aims to compare the recognition performance of seven widely used machine learning algorithms, including traditional and deep learning ones, over a small dataset. The experimental results reveal that traditional machine learning algorithms such as Support Vector Machine and Random Forest outperform deep learning algorithms such as Multi-Layer Perceptron and Convolutional Neural Network.
Student Author(s): Samriddhi Gautam
Faculty Author(s): Dr. Junxiu Zhou, Dr. Yangyang Tao
Author Department(s): School of Computing and Analytics
Title: Text Mining on Face Recognition Literatures
Abstract: Facial recognition related literatures have been widely researched and discussed in the recent years, since it's one of the major biometric tools, which is associated with Artificial Intelligence. However, organizing, understanding, and extracting useful information from these articles could be very time-consuming and labor-intensive. Therefore, we applied text mining technology to obtain, organize, classify, summarize literatures related to facial recognition. Specifically, we first collected a face recognition abstract (FRA) dataset from the PubMed database. Then, we used topic modeling Latent Dirichlet Allocation (LDA), which is one of the most preferred topic modeling methods, to perform text mining on the collected FRA dataset. Finally, results generated by LDA were analyzed to provide an overview of the FRA dataset.
