RSS 1.0Doctoral Degree Granting Institutions
Browse by
All campuses are research institutions and offer bachelor's, master's and doctoral degrees through a broad range of academic programs. Undergraduate research and honors programs provide students with opportunities to interact one-on-one with renowned national and international faculty.
Sub-communities within this community
Recent Submissions
-
Feasibility and applicability of a clinical assessment of both the ON and OFF pathways in patients with glaucoma and controls."Purpose: To assess the feasibility and clinical utility of a head-mounted, On/Off perimetry test and to investigate the effect of early to moderate glaucoma on reaction time and accuracy to ON and OFF perimetric stimuli. Methods: We tested one eye each of 9 patients with early to moderate primary open angle glaucoma (mean = 71.88 years, std = 5.17), 9 visually-normal control patients of a similar age (mean = 63.88 years, std = 5.17 ) , and 9 visually-normal optometry students (ages 22-25 years). We used a head mounted display equipped with an eye tracker (HTC VIVE embedded Tobii). Custom software (Unity, version 2017) was used to create the stimuli and a library provided by Tobii Pro was used to measure eye movements at 120 Hz. Stimulus size changed as a function of eccentricity using a power law relationship: stimulus size= minimum scale*(eccentricity/5)^α. Eye movements were restricted to a central circle with a 2.5 degree radius. Stimulus contrast was initially set to 100%. A single test comprised of 579 trials, including 51 catch trials, presented at 90 different positions in the visual field. Each test location was repeated 3 times for both light and dark stimuli, with 6 repeats in each of two blind spot positions. Results: Our results demonstrate asymmetry between the two achromatic visual transduction pathways. These results support previous findings that dark targets elicit a faster and more accurate response than light targets, when presented on binary background noise. Our results extend previous work by demonstrating that the two pathways remain asymmetrical in eccentricities up to 30 degree from fixation. We also show that the relationship between the percentage of correct responses for ON pathway and OFF pathway stimuli follows a power function, wherein glaucoma and controls overlap (R2=0.842) . This overlap decreases when we quantify only the subthreshold (unseen) increment targets in a linear relationship (R2=0.7074). All controls had less than 12% of subthreshold increment targets whereas the percentage of subthreshold targets was higher for 75% of the glaucoma subjects, even in early stages of the disease. CONCLUSION We have demonstrated that ON/OFF perimetry is feasible in a VR environment and confirmed an asymmetry between the ON and OFF pathways in patients with glaucoma and control patients in both central and peripheral visual fields. We measured on-pathway deficits in patients with limited loss of visual sensitivity which may improve detection of early disease. Future work will focus on optimizing stimulus parameters to improve the sensitivity and specificity of this test."
-
Effects Of Correcting Fixation Disparity On Digital Eye StrainDigital Eye Strain (DES) is a widespread and highly prevalent condition whose incidence appears to be rising during the present pandemic. It comprises a range of visual and ocular symptoms which occur after viewing a digital screen for an extended period of time. Previous work from our laboratory has shown the magnitude of fixation disparity to be the only clinical parameter that is significantly correlated with DES symptoms. Therefore, this study sought to determine whether correcting the underlying fixation disparity will significantly reduce DES symptoms. Thirty young, visually-normal students were required to read randomly generated words from a digital tablet device for 20 minutes. Three different trials were performed, with the subject wearing either: (i) the prism that corrected their fixation disparity, (ii) the same magnitude of prism as for condition (i) but with the opposite base direction or (iii) a near addition lens that corrected the fixation disparity. Immediately after the reading task, subjects rated their ocular and visual symptoms on a questionnaire. There was no significant difference between the mean symptom scores for the three conditions. However, this may be due, in part, to the small number of subjects encountered with large values of fixation disparity. Future studies should further examine the range of oculomotor responses associated with DES in order to provide appropriate treatment options.
-
Generating Facial Character: A Systematic Method Accumulating Expressive HistoriesThe author presents a method to simulate facial character development by accumulating an expressive history onto a face. The model analytically combines facial features from Paul Ekman’s seven universal facial expressions using a simple Markov chain algorithm. The output is a series of 3D digital faces created in Blender with Python. The results show that systematically imprinting features from emotional expressions onto a neutral face transforms it into one with distinct character. This method could be applied to creative works that depend on character creation, ranging from figurative sculpture to game design, and allows the creator to incorporate chance into the creative process. The author demonstrates the method’s application to sculpture with ceramic casts of generated faces.
-
Library renovation: From setbacks to success.The journey started at a 2016 meeting of the college’s Learning Resources Committee, which I chair as Library Director, A student representative casually commented that our Library looked “dated”. Not that we actually WERE dated, we have electronic resources, 24/7 remote access, printers, scanners, wireless, and a state-of-the-art Library Management System. However, through the eyes of our young student, the Library looked “like something from the 1970s.” Anyone who has lived through that decade’s will know that was not meant as a compliment. The student’s criticism prompted the Dean of Academic Affairs, who sat in on the meeting, to ask when the Library had had its last “facelift.” That was an easy one to answer - as far as I knew…never. And I’ve been here over 20 years. Just like that, a renovation was born. How hard could this be? I would query the students on what a suitable update might look like, get a budget, buy some new furniture, and go back to the business of being a librarian rather than an interior designer. Spoiler alert: not so fast.
-
Crowdsourcing Image Extraction and Annotation: Software Development and Case StudyWe describe the development of web-based software that facilitates large-scale, crowdsourced image extraction and annotation within image-heavy corpora that are of interest to the digital humanities. An application of this software is then detailed and evaluated through a case study where it was deployed within Amazon Mechanical Turk to extract and annotate faces from the archives of Time magazine. Annotation labels included categories such as age, gender, and race that were subsequently used to train machine learning models. The systemization of our crowdsourced data collection and worker quality verification procedures are detailed within this case study. We outline a data verification methodology that used validation images and required only two annotations per image to produce high-fidelity data that has comparable results to methods using five annotations per image. Finally, we provide instructions for customizing our software to meet the needs for other studies, with the goal of offering this resource to researchers undertaking the analysis of objects within other image-heavy archives.
-
Faces extracted from Time Magazine 1923-2014We present metadata of labeled faces extracted from a Time magazine archive that contains 3,389 issues ranging from 1923 to 2012. The data we are publishing consists of three subsets: Dataset 1) the gender labels and image characteristics for each of the 327,322 faces that were automatically-extracted from the entire Time archive, Dataset 2) a subset of 8,789 faces from a sample of 100 issues that were labeled by Amazon Mechanical Turk (AMT) workers according to ten dimensions (including gender) and used as training data to produce Dataset 1, and Dataset 3) the raw data collected from the AMT workers before being processed to produce Dataset 2.
-
What’s in a Face? Gender Representation of Faces in Time, 1940s-1990sWe extracted 327,322 faces from an archive of Time magazine containing 3,389 issues dating from 1923 to 2014, classified the gender of each extracted face, and discovered that the proportion of female faces contained within this archive varied in interesting ways over time. The proportion of female faces first peaked in the mid-to-late 1940s. This was followed by a dip lasting from the mid-1950s to the early 1960s. The 1970s saw another peak followed by a dip over the course of the 1980s. Finally, we see a slow and steady rise in the proportion of female faces from the early 1990s onwards. In this paper, we seek to make sense of these variations through an interdisciplinary framework drawing on psychology, visual studies (in particular, photography theory), and history. Through a close reading of our Time archive from the 1940s through the 1990s, we conclude that the visual representation of women in Time magazine correlates with attitudes toward women in both the historical context of the era and the textual content of the magazine.
-
Color Transparency: Geometry, Motion, Color, Scission, and InductionObjects that pass light through are considered transparent, and we generally expect that the light coming out will match the perceived color of the object. However, when the object is placed on a colored surface, the light coming back to our eyes becomes a composite of surface, illumination, and transparency properties. Despite that, we can often perceive separate overlaid and overlaying layers differing in colors. How neurons separate the information to extract the transparent layer remains unknown, but physical characteristics of transparent filters generate geometrical and color features in retinal images which could provide cues for separating layers. We estimated the relative importance of such cues in a perceptual scale for transparency, using stimuli in which X or T-junctions, different relative motions, and color consistency, cooperated or competed in forced-preference psychophysics experiments. Maximum-likelihood Thurstone scaling revealed some new results: moving X-junctions increased transparency compared to static X-junctions, but moving T-junctions decreased transparency compared to static T-junctions by creating the percept of an opaque patch. However, if the motion of a filter uncovered a dynamically changing but stationary pattern, sharing common fate with the surround but forming T-junctions, the probability of seeing transparency was almost as high as for moving X-junctions, despite the stimulus being physically improbable. In addition, geometric cues overrode color inconsistency to a great degree. Finally, a linear model of transparency perception as a function of relative motions between filter, overlay, and surround layers, contour continuation, and color consistency, quantified a hierarchy of latent influences on when the filter is seen as a separate transparent layer. Previous measurements of color scission have limitations. The color adjustment to match the target color is relatively accurate but time consuming and suffers from long time adaptation bias. Force-choice judgment is quick and free from adaptation effect, but the selection of choices can be the source of bias. We examine the observers’ ability to estimate filter color with transparency with our improved method: we ask observers to make a judgement of the transparency region being red or green (or, blue/yellow). By doing this we found the neutral point of the filter that the observers think colorless. The result showed that, in color consistent conditions, though biased by the background or individual preference, the observers’ measured neutral filter settings were close to the colorless filter, showing relatively good color scission. In the color inconsistent conditions, the observers matched the overlaid region to a neutral color, as if the observers were attributing the average color of the overlaid region completely to the transparency. Veridicality of scission varied little in the color consistent conditions, despite the large variation in degree of perceived transparency. An exception to the rule that only one color is seen at every retinotopic location happens when a bounded colored transparency or spotlight is seen on a differently colored surface. Despite the spectrum of the light from each retinotopic location being an inextricable multiplication of illumination, transmission, and reflectance spectra, we seem to be able to scission the information into background and transparency/spotlight colors. Visual cues to separating overlay and overlaid layers have been enumerated, but neural mechanisms that extract veridical colors for overlays have not been identified. Here, we demonstrate that spatial induction contributes to color scission by shifting the color of the overlay toward the actual color of the filter. By alternating filter and illumination spectra, we present naturalistic simulations where isomeric disks appear to be covered by filters/spotlights of near veridical colors, depending solely on the surrounding illumination.
-
Face Recognition and Emotion IdentificationWhile face recognition has been around in one form or another since the 1960s, recent technological developments have led to a wide proliferation of this technology. This technology is no longer seen as something out of science fiction movies like Minority Report. With the release of the iPhone X, millions of people now literally have face recognition technology in the palms of their hands, protecting their data and personal information. While mobile phone access control might be the most recognizable way face recognition is being used, it is being employed for a wide range of use cases including preventing crime, protecting events and making air travel more convenient. This project focuses on various advanced Python libraries to improve the face recognition accuracy such as OpenCV, Sklearn, face_recognition. The project understands the data and model, train it for further usage. The real time videos are considered for evaluating the results. Further the project glances the emotion recognition algorithms using CV2, Seaborn. The areas of the human faces are highlighted according to different emotions. The large data sets (fer2013, Olivetti faces) are used for training and testing the data sets. PCA, leave one out cross validation, grid search CV, machine learning pipelines, CNN models are used to estimate and increase the accuracy. The project is executed in Anaconda environment Jupyter Notebook. As the data sets are huge Google Collaboratory is used for execution.
-
sxRNA Switches: Hypothesis Through Automated Design Via a Genetic Algorithm ApproachThe following document is meant to represent an overview of my work on structurally interacting RNA (sxRNA), which has already resulted in three publications with another two in preparation. Where appropriate, some text and data from these publications have been reproduced here. Ribonucleic Acid (RNA) is one of the fundamental macromolecules present in living systems. It can be found in all cells as varying length polymer chains composed of four primary bases (adenine, cytosine, guanine, uracil) capable of numerous modifications. Though generally characterized as an information carrier, RNA is a versatile molecule that exhibits both intra and inter-strand base pairing to form complex structures. Similar to protein, the particular shape of an RNA structure in combination with some degree of sequence specificity, can dictate its function (RNA binding protein recognition sites, ribozyme activity, aptameric affinity, etc.). Structurally interacting RNA (sxRNA) is a molecular switch technology that exploits predictable intermolecular RNA base pairing to form an otherwise absent functional structure in one RNA strand when it interacts with a specific, targeted second strand. Originally proposed as a potential regulatory mechanism in natural systems, we used characteristics of predicted pairings in that context to engineer purely synthetic sxRNA switches that have been successfully tested. There are many non-coding RNAs associated with pathological conditions, the ability to use these as triggers for sxRNA opens the door to potential applications ranging from diagnostics to therapeutics. Furthermore, other prospective triggers (including those synthetically designed) may allow use of the technology as a molecular tool for a variety of purposes including as an alternative to antibiotic selection in cell line development. The typical trigger sequences targeted by sxRNA switches are at least 20 bases in length. Combinatorial options with regard to structure positioning and base composition produce an enormous number of potential sxRNA sequences for any given target. Exhaustively examining these for feasible candidates (i.e., analyzing predicted interactions with unintended targets) is computationally impossible with current systems. Evolutionary computing is a subfield of artificial intelligence (AI) that has been inspired by biology. Genetic algorithms are a type of evolutionary algorithm and apply operators (such as recombination and mutation) to find candidate solutions to an optimization problem. The presented dissertation will describe the original sxRNA research as well as the development and testing of a genetic algorithm that automates the production of new sxRNA switch candidates. This algorithm takes into consideration factors that were previously impossible to account for in manual designs.
-
Multicyclic Loss for Multidomain Image-to-Image TranslationGANs developed to Translate an Image’s style between different domains often only care about the initial translation, and not the ability to further translate upon an image This can cause issues where, if one would want to generate upon an image and then further on, change that image even more that person may come into issues. This creates a ”gap” between the base images and the generated images, and in this paper a Multicyclic Loss is presented, where the Neural Network also trains on further translations to images that were already translated. iv
