In recent times there has been an explosion of online user-generated video content. This has generated significant research interest in video analytics. Human users understand videos based on high-level semantic concepts. However, most of the current research in video analytics are driven by low-level features and descriptors, which often lack semantic interpretation. Existing attempts in semantic video analytics are specialized and require additional resources like movie scripts, which are not available for most user-generated videos. There are no general purpose approaches to understanding videos through semantic concepts. In this thesis we attempt to bridge this gap. We view videos as collections of entities which are semantic visual concepts like the persons in a movie, or cars in a F1 race video. We focus on two fundamental tasks in Video Understanding, namely summarization and scene- discovery. Entity-driven Video Summarization and Entity-driven Scene discovery are important open problems. They are challenging due to the spatio-temporal nature of videos, and also due to lack of apriori information about entities. We use Bayesian nonparametric methods to solve these problems. In the absence of external resources like scripts we utilize fundamental structural properties like temporal coherence in videos- which means that adjacent frames should contain the same set of entities and have similar visual features. There have been no focussed attempts to model this important property. This thesis makes several contributions in Computer Vision and Bayesian nonparametrics by addressing Entity-driven Video Understanding through temporal coherence modeling. Temporal Coherence in videos is observed across its frames at the level of features/descriptors, as also at semantic level. We start with an attempt to model TC at the level of features/descriptors. A tracklet is a spatio-temporal fragment of a video- a set of spatial regions in a short sequence (5-20) of consecutive frames, each of which enclose a particular entity. We attempt to find a representation of tracklets to aid tracking of entities. We explore region descriptors like Covari- ance Matrices of spatial features in individual frames. Due to temporal coherence, such matrices from corresponding spatial regions in successive frames have nearly identical eigenvectors. We utilize this property to model a tracklet using a covariance matrix, and use it for region-based entity tracking. We propose a new method to estimate such a matrix. Our method is found to be much more efficient and effective than alternative covariance-based methods for entity tracking. Next, we move to modeling temporal coherence at a semantic level, with special emphasis on videos of movies and TV-series episodes. Each tracklet is associated with an entity (say a particular person). Spatio-temporally close but non-overlapping tracklets are likely to belong to the same entity, while tracklets that overlap in time can never belong to the same entity. Our aim is to cluster the tracklets based on the entities associated with them, with the goal of discovering the entities in a video along with all their occurrences. We argue that Bayesian Nonparametrics is the most convenient way for this task. We propose a temporally coherent version of Chinese Restaurant Process (TC-CRP) that can encode such constraints easily, and results in discovery of pure clusters of tracklets, and also filter out tracklets resulting from false detections. TC-CRP shows excellent performance on person discovery from TV-series videos. We also discuss semantic video summarization, based on entity discovery. Next, we consider entity-driven temporal segmentation of a video into scenes, where each scene is characterized by the entities present in it. This is a novel application, as existing work on temporal segmentation have focussed on low-level features of frames, rather than entities. We propose EntScene: a generative model for videos based on entities and scenes, and propose an inference algorithm based on Blocked Gibbs Sampling, for simultaneous entity discovery and scene discovery. We compare it to alternative inference algorithms, and show significant improvements in terms of segmentatio and scene discovery. Video representation by low-rank matrix has gained popularity recently, and has been used for various tasks in Computer Vision. In such a representation, each column corresponds to a frame or a single detection. Such matrices are likely to have contiguous sets of identical columns due to temporal coherence, and hence they should be low-rank. However, we discover that none of the existing low-rank matrix recovery algorithms are able to preserve such structures. We study regularizers to encourage these structures for low-rank matrix recovery through convex optimization, but note that TC-CRP-like Bayesian modeling is better for enforcing them. We then focus our attention on modeling temporal coherence in hierarchically grouped sequential data, such as word-tokens grouped into sentences, paragraphs, documents etc in a text corpus. We attempt Bayesian modeling for such data, with application to multi-layer segmentation. We first make a detailed study of existing models for such data. We present a taxonomy for such models called Degree-of-Sharing (DoS), based on how various mixture components are shared by the groups of data in these models. We come up with Layered Dirichlet Process which generalizes Hierarchical Dirichlet Process to multiple layers, and can also handle sequential information easily through Markovian approach. This is applied to hierarchical co-segmentation of a set of news transcripts- into broad categories (like politics, sports etc) and individual stories. We also propose a explicit-duration (semi-Markov) approach for this purpose, and provide an efficient inference algorithm for this. We also discuss generative processes for distribution matrices, where each column is a probability distribution. For this we discuss an application: to infer the correct answers to questions on online answering forums from opinions provided by different users.

Parent-child interactions help set the foundation for a child's development. It is therefore important to investigate the relative efficiency and efficacy of procedures used to train them. One procedure that researchers continue to explore is video modeling. The current study evaluated the effect of a video model that displayed favorable parent-child interactions and a modified model with embedded instructions to determine if the introduction of either of these models would alter parent-child interactions. Both models were presented alone without supplemental guidance. Three families were involved in the study. The results showed no systematic change across families or conditions as a result of video viewing and are discussed within context of the needs of the parent, adequate stimulus control, community to support behavior change, measurement sensitivity, and influence of methodology. This study provided a great baseline for future studies to explore the necessary components to create an effective video model.

Young adults with Autism Spectrum Disorder (ASD) face unique social skills challenges as they transition into independent living environments and seek fulfilling relationships within their communities. Research has focused on social education and interventions for children with autism, while transitioning young adults with ASD have received insufficient attention. The purpose of this multisite case study was to explore perceptions of school personnel related to the use of visual support system (VSS) technology and enhancement of social skillsets for young adults with ASD. Information processing theory and social learning theory provided the research framework. Research questions addressed perceptions related to the utility of VSS technology and social skills teaching strategies. Interviews were conducted with 11 special education administrators, teachers, and intervention specialists from 3 different programs in the United States. Data from interviews and field notes were analyzed using open, axial, and selective coding; themes such as social skills, video-modeling, learning strategies, use of visual technology, and cognition emerged. Participants indicated that exploring cognitive learning strategies underpinned with VSS technology provided alternative methods to teach social skills in classroom settings. They identified the need for more funding for VSS technologies for all learners. Implications for social change include that social skills and critical thinking skills can be enhanced by learning through the use of VSS technology. Empowering young adults with ASD to participate with greater confidence in learning situations and in social situations will support their efforts to be more comfortable and to interact more appropriately in work and community interactions.

Many researchers feel that teacher preparation programs are not doing enough to prepare teachers to effectively use technology. The result is a plethora of teachers who may know the basic functions of different programs, but who are unprepared to integrate these skills into their teaching. One method used by a few preservice programs, including BYU's, is the use of modeling sessions, otherwise referred to as live modeling. In these modeling sessions, the instructor models for the preservice teachers how a K-12 teacher could teach with technology, while the preservice teachers participate as if they were K-12 students. This thesis is a qualitative investigation of how this method of live modeling has impacted students, according to the perceptions of a sample of former students of the course. This project also has a practical focus of identifying strategies for improving modeling, and pitfalls that may indicate when modeling is not as effective. Overall, this study found that modeling was perceived by most students to be effective at teaching technology skills and ideas for integrating technology as teachers. However, there were some students who struggled to abstract principles from the modeling that could help them as teachers. In other words, they struggled to cognitively transfer the learning from the context of the modeling session to their own teaching contexts. In this research I identify five main contextual breakdowns that often occurred among students in the course. These were breakdowns, or differences, between the modeled context and the students' actual contexts that were sufficiently large enough to disrupt the students' abilities to cognitively transfer the learning. By adapting the live modeling method to more specifically address unique students' needs and contexts, then the cognitive transfer of learning should be easier and the method could be a strong tool for training preservice teachers to use technology in their own teaching.

Over the last decade, the inclusion of students with disabilities (SWD) in the general education classroom has increased. Currently, 60% of SWD spend 80% or more of their school day in the general education classroom (U.S. Department of Education, 2013). This includes students with autism spectrum disorders (ASD), a developmental disability characterized by impairments in behavior, language, and social skills (American Psychological Association, 2013). Many of these SWD struggle with mathematics in the elementary grades; fewer than 20% of SWD are proficient in mathematics when they begin middle school, compared to 45% of their peers without disabilities. Furthermore, 83% of SWD are performing at the basic or below basic level in mathematics in the fourth grade (U.S. Department of Education, 2013). As the rate of ASD continues to increase (Centers for Disease Control, 2013), the number of students with this disability who are included in the general education classroom also continues to rise. These SWD and students with ASD are expected to meet the same rigorous mathematics standards as their peers without disabilities. This study was an attempt to address the unique needs of SWD and students with ASD by combining practices rooted in the literature, strategy instruction and video modeling. The purpose of this study was to determine the effects of an intervention on the ability of students with and without disabilities in inclusive fourth and fifth grade classrooms to solve word problems in mathematics. The intervention package was comprised of a graphic organizer, the K-N-W-S, video models of the researcher teaching the strategy to a student avatar from a virtual simulated classroom, TeachLivE, and daily word problems for students to practice the strategy. The researcher used a quasi-experimental group design with a treatment and a control group to determine the impact of the intervention. Students were assessed on their performance via a pretest and posttest. Analyses of data were conducted on individual test items to assess patterns in performance by mathematical word problem type. The effects of the intervention on SWD, students with ASD, and students without disabilities varied widely between groups as well as amongst individual students, indicating a need for further studies on the effects of mathematics strategy instruction on students with varying needs and abilities.

This research examined the effects of self-regulated strategy development (SRSD), a cognitive strategy instructional method, on opinion writing by third grade students with learning disabilities. A video self-modeling (VSM) component was added to the SRSD method. A multiple probe across participants, single-subject design was used to determine the effectiveness of the SRSD instructional strategy, (POW + TREE), in combination with video self-modeling. Data from various components of writing, including essay elements, length of responses, time spent writing, and overall writing quality, were collected and assessed to determine the effectiveness of the intervention. All students who received the intervention improved their overall writing performance on opinion essays as measured by the number of opinion essay elements, including topic sentence, reasons, examples, and ending. During the maintenance phase of the intervention, students who received a VSM booster session increased their total number of opinion essay elements back to mastery levels.

No description available.

Individuals with autism spectrum disorder (ASD) display deficits in communication and social interaction that can impact their ability to function in daily environments. To remediate these deficits, it is critical for professionals to use effective interventions. While there are many evidence-based practices (EBPs) identified for ASD (e.g., video modeling), the adoption of these EBPs may not occur automatically. Existing research suggests professionals have a generally favorable impression of video modeling. However, little research has examined opinions and applied use of video modeling, which was the purpose of the present study. Using survey methodology, data were collected from 510 professionals in various disciplines (e.g., special educators, speech-language pathologists [SLPs], and behavior analysts [BCBAs]). Data were analyzed primarily via factor analysis and multiple regression. Factor analysis was used to examine the underlying structure of the instrument, revealing two predominant factors: (1) interest in and (2) perceived accessibility of video modeling. Multiple regression was used to examine which demographic characteristics (e.g., age and years of experience) were associated with each factor. Results indicated that BCBAs and SLPs perceived video modeling as more accessible. In terms of interest, professionals who worked with preschool-aged students, who worked in a suburban location, and who had an extended family member with ASD showed higher interest in video modeling. Implications for practice and future research are discussed.

Discrete trial instruction (DTI) and naturalistic developmental behavioral interventions (NDBIs) are often incorporated into early intensive behavioral interventions for young children with autism. Recent advances in staff training methods have demonstrated that self-instructional manuals, video models, and computer-based training are effective and efficient ways to improve staff implementation of these teaching strategies however research in this area is limited. The current evaluation assessed the effects of a computer-based training package including self-instructional manuals with embedded video models on direct-care staff’s implementation of DTI and an NDBI. All participants’ DTI teaching fidelity increased during role-plays with an adult and with a child with autism and all participants increased teaching fidelity across untrained instructional programs. In addition, moderate improvement was demonstrated following NDBI training on the use of correct prompts, environmental arrangements, and response interaction. Together, these results indicate that therapists are able to acquire a large number of skills using two teaching techniques, DTI and NDBI, following brief computer-based training.

No description available.