UMD Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/3
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a given thesis/dissertation in DRUM.
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
4 results
Search Results
Item Supervision and Data Dynamics in Vision Across Recognition and Generation Landscapes(2024) Suri, Saksham; Shrivastava, Abhinav; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This thesis looks at visual perception through the lens of supervision and data dynamics across recognition and generation landscapes. Generative and discriminative modeling form important pillars in computer vision. Depending on the task techniques to better learn and utilize the data and labels can change. Through this work we investigate different tasks along this landscape focusing on different supervision strategies, highlighting pitfalls in current approaches and propose modified architectures and losses to utilize the data better under different settings. On the recognition side we start by analyzing Vision Transformers (ViTs) through a comprehensive analysis under varied supervision paradigms. We look at a mix of explicit supervision, contrastive self-supervision, and reconstructive self-supervision by delving into attention mechanisms and learned representations. We then look at a more specific case of supervision geared towards object detection which is called sparse supervision where their are missing annotations. We propose to utilize self and semi-supervised techniques to solve this task. Finally, we also explore a discovery style framework with applications on GAN generated image detection. Unlike sparse supervision discussed earlier, this scenario handles the case where are test time we have an unknown number of new classes. We were the first work proposing this problem where instead of just identifying synthetic images, we also try to group them based on their generation source. The exploration of Generative Adversarial Networks (GANs) in an open-world scenario uncovers the intricacies of learning with limited supervision for discovery style problems. On the generation side we delve into different supervision strategies involving decomposing and decoupling representations. In the first work we tackle the problem of paired Image-to-Image (I2I) translation by decomposing supervision into reconstruction and residuals and highlight issues with traditional training approaches. We then look at generating talking head videos through two different kinds of supervision, video and audio. For driving the generation using a video we look at decoupling representations for the task of few-shot talking-head synthesis where the supervision is provided using only a few samples (shots). For this task we factorize the representation into spatial and style components which helps the learning. To supervise the generation additionally through audio, we look at multimodal supervision for lip-synchronized talking head generation. For this we incorporate audio and video modalities to synthesize lifelike talking-heads which can work even in in-the-wild scenarios. In the last part we showcase two works which link our experiences from generation and recognition where we explore generative modeling to improve recognition models. The first work here utilizes the advancements in diffusion based image generation models to improve recognition models. Given the high fidelity and control of generation which diffusion models have brought, we utilize synthetic data from these models and create a suitable pipeline to utilize this data effectively to improve detection and segmentation performance. As a follow up to our ViT analysis we also propose a new technique to utilize off the shelf pretrained ViTs and generate high resolution features using a learnt lightweight feature transform. These high resolution features are especially effective for dense tasks like correspondence, segmentation, detection and object discovery.Item SUPERVISORY STYLES, SUPERVISION OUTCOME AND COUNSELOR SELF-EFFICACY OF ADDICTION TREATMENT PROFESSIONALS(2012) Berger, Gisela P.; Holcomb-McCoy, Cheryl; Counseling and Personnel Services; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Due to the paucity of research in the addictions treatment field regarding supervision, this research study sought to examine the style, type and outcome of supervision of addictions treatment professionals and how it relates to supervisee self-efficacy. Thus, the primary purpose of this study is to quantify the addictions treatment professional's supervision outcome in relation to other variables including supervisory style, supervisory type, and demographic variables. The research questions of the present study are as follows: 1. What do addictions treatment counselors perceive as the overall supervisory style of their clinical supervisors? 2. Does the supervisory style of clinical supervisors of addictions treatment counselors predict the outcome of supervision? 3. What type of delivery style of supervision do addictions treatment counselors receive? 4. Is there a significant relationship between supervision outcome and specific demographic characteristics of addictions treatment professionals (e.g., age, years of experience, recovery status, gender, formal education)? 5. What is the relationship between perceived addictions treatment counselor self-efficacy and perceptions of supervision outcome? 6. Can supervision outcome be predicted from addiction treatment counselor perceived supervisory style and perceived self-efficacy? This study discussed the definition of supervision, types and techniques of supervision, models, and mechanisms. The supervisory relationship, supervision outcome, and supervision effectiveness were discussed in relation to other areas of mental health counseling. When discussing these areas in terms of the addictions treatment field, the lack of empirically valid research was noted. The many different types of supervision (e.g., individual, group, peer, etc.) and different styles were also discussed. Though styles have been variously defined, this research study used the supervisory styles as defined by Friedlander & Ward (1984) which include the Attractive, Interpersonally Sensitive, and Task Oriented subscales. The methodology included a pilot study to refine the instruments that were ultimately used for this study. Several changes were made to the original instruments and these changes focused mainly on defining terms used in rating supervisors. The three instruments that were used include the Supervisory Styles Inventory (SSI; Friedlander & Ward, 1984), The Supervision Outcome Questionnaire (SOQ; Worthington and Roehlke (1979) and the Counselor Self-Efficacy Scale (COSES; Melchert, et. al., 1996). A demographic questionnaire was included to gather information regarding demographic variables and supervisory style. The results indicate that supervisees have a fairly positive overall perception of the supervisory style of their supervisors. Supervisory style was found to be predictive of supervision outcome with a positive linear relationship. The delivery style of supervision for this sample was noted. There were no demographic variables that predicted type of supervision. Counselor self-efficacy was found to be not statistically significant relative to supervision outcome. A linear relationship was found between supervision style, counselor self-efficacy, and supervision outcome such that when overall supervision style increases and self-efficacy decreases slightly, then supervision outcome can be predicted. Finally, the implications for training of supervisors in the addictions treatment field was discussed as well as the implications for future research. The study's limitations were also discussedItem Trainees' Use of Supervision for Clinical Work with Lesbian, Gay, and Bisexual Clients: A Qualitative Study(2010) McGann, Kevin; Hoffman, Mary Ann; Counseling and Personnel Services; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Most studies considering lesbian, gay, and bisexual (LGB) clients and their therapists have primarily focused on the clients' experience in therapy (Israel, Walther, Gorcheva, & Sulzner, 2007; Liddle, 1996) or on the therapist's experience of the client (Bieschke & Matthews, 1996; Garnets, Hancock, Cochran, Goodchilds, & Peplau, 1991). However, the role that clinical supervision plays in therapists' development in working with LGB clients is rarely studied. Not enough is known about how supervision is perceived from the perspective of the supervisee who is developing skills in working with LGB clients. The current study examined 12 interviews with randomly selected predoctoral interns at APA-accredited counseling centers around the country, to explore how they made use of the clinical supervision they received for their work with LGB clients. The single previous qualitative investigation of this topic (Burkard, Knox, Hess, & Shultz, 2009) examined interviews with LGB advanced doctoral students. The current study extends the investigation of this topic by interviewing six heterosexual-identified trainees in addition to six trainees who identified as LGB or queer (Q), and by investigating a more geographically heterogeneous sample. The interviews explored various aspects of the supervision experience, including trainees' expectations of their supervisor for supervision of their work with their LGB client, the contributions of trainees and their supervisors to the supervision process, and the impact of supervision on work with the LGB client and other clients. The data were analyzed using Consensual Qualitative Research (CQR; Hill, Knox, Thompson, Williams, Hess, & Ladany, 2005; Hill, Thompson, & Williams, 1997). All participants valued their supervision relationship, and found their supervisors helpful in assisting them in their therapeutic work with their LGB client. Trainees typically experienced their supervisors as multiculturally sensitive, and some felt that their supervisors helped them with LGB-specific interventions and case conceptualizations. Some differences between how heterosexual and LGB-identified trainees used supervision for their work with their LGB clients. All participants reported gains from their supervision experience with their LGB client that positively affected their work with other clients, regardless of these clients' sexual orientation.Item Is the Influence of Unsupervised Time with Peers on Delinquency Conditioned by the Deviance of One's Peers?(2006-04-28) Bauer, Erin Louise; Gottfredson, Denise C.; Criminology and Criminal Justice; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Drawing on applications of opportunity theory to juvenile delinquency, this study examines the relationship between time expenditure and delinquency. This research relies on self-report data from the Maryland After School Opportunity Fund Program (MASOFP) with a sample of 817 adolescents. The conclusions of this study are: (1) Supervised time without peers is not conducive to delinquency, while supervised time with peers is conducive to delinquency. (2) Unsupervised time with and without peers is conducive to delinquency. (3) Unsupervised time with peers is most conducive to delinquency. (4) The relationship between unsupervised time with peers and each category of delinquency (property, violent, and substance use) is similar. (5) The relationship between unsupervised time with peers and each violent delinquency offense is similar. (6) The relationship between unsupervised time with peers and delinquency is not conditioned by the deviance of one's peers.