Wanli Xing’s research while affiliated with University of Florida and other places

The current study explores the use of computer vision and artificial intelligence (AI) methods for analyzing 360-degree spherical video-based virtual reality (SVVR) data. The study aimed to explore the potential of AI, computer vision, and machine learning methods (including entropy analysis, Markov chain analysis, and sequential pattern mining), in extracting salient information from SVVR video data. The research questions focused on differences and distinguishing characteristics of autistic and neurotypical usage characteristics in terms of behavior sequences, object associations, and common patterns, and the extent to which the predictability and variability of findings might distinguish the two participant groups and provide provisional insights into the dynamics of their usage behaviors. Findings from entropy analysis suggest the neurotypical group showed greater homogeneity and predictability, and the autistic group displayed significant heterogeneity and variability in behavior. Results from the Markov Chains analysis revealed distinct engagement patterns, with autistic participants exhibiting a wide range of transition probabilities, suggesting varied SVVR engagement strategies, and with the neurotypical group demonstrating more predictable behaviors. Sequential pattern mining results indicated that the autistic group engaged with a broader spectrum of classes within the SVVR environment, hinting at their attraction to a diverse set of stimuli. This research provides a preliminary foundation for future studies in this area, as well as practical implications for designing effective SVVR learning interventions for autistic individuals. 1. Background The current article presents an exploratory study that investigates the use of computer vision and AI methods for analyzing 360-degree spherical video data, also referred to as spherical, video-based virtual reality, or SVVR (Chien et al., 2020). This study provides valuable insights for researchers who are interested in using AI, computer vision, and machine learning for analyzing immersive learning intervention data collected within autistic and neurotypical groups. The purpose of the study is to demystify the methods and processes used and provide a clear understanding of how these technologies can be applied in the field of immersive learning interventions. This article will be of value to researchers who are looking to explore the use of AI and computer vision for analyzing 360-degree SVVR data, providing a starting point for further research in this field and shedding light on the potential applications and benefits of these technologies. Conducting qualitative analysis necessitates substantial allocation of resources and time, which has been well-established in the research community (Smith, 2018; Patton, 2015; Heath et al., 2010). Video analysis, in particular, presents a formidable challenge due to its intricate nature, often demanding specialized technology and labor-intensive manual processes (Atkinson, 2007; Knoblauch et al., 2008). These processes can involve procedures such as: (1) crafting descriptive narratives of actors and activities (Laurier, 2019), (2)

One of the primary obstacles to improving middle school math achievement is lack of equitable access to high-quality learning opportunities. Human delivery of high-dosage tutoring can bring significant learning gains, but students, particularly economically disadvantaged students, have limited access to well-trained tutors. Augmenting human tutor abilities through the use of artificial intelligence (AI) technology is one way to scale up access to tutors without compromising learning quality. This workshop aims to highlight the challenges and opportunities of AI-in-the-loop math tutoring and encourage discourse in the AIED community to develop human-AI hybrid tutoring and teaching systems. We invite papers that provide clearer understanding and support the progress of human and AI-assisted personalized learning technologies. The structure of this full-day hybrid workshop will include presentations of accepted papers, small or whole group discussion, and a panel discussion focusing on common themes related to research and application, key takeaways, and findings imperative to increasing middle school math learning.KeywordsTutoringPersonalized learningAI-assisted tutoring

Computer vision (CV) is a subset of artificial intelligence (AI) that focuses on enabling computers to detect and understand objects from visual stimuli and multimedia. With advances in computational power and open-source libraries, more educators and instructional designers are seeking to capitalize on the perceived benefits of CV. This work-in-progress paper reports the CV approach our research team has developed to explore the usage patterns of neurodiverse learners within a 360-degree spherical video-based virtual reality system. Index terms-computer vision, SVVR, artificial intelligence, 360-video, autism I. BACKGROUND & RELATED WORK Research on the use of artificial intelligence (AI) in education has been ongoing for over 30 years [1]. In this time, researchers have explored widely the potential of AI to facilitate the teaching and learning process and to simulate human intelligence, with the goal of facilitating inferences, judgments, and/or predictions in educational settings [2]. Recent computational advancements have led to a renewed interest in using AI technologies in learning contexts, a growing number of studies are underway to explore the tremendous potential of this technology [3]. Among AI approaches, computer vision (CV) has emerged as a particularly relevant technology for assisting instructors in evaluating student engagement and evaluating instructional materials. CV is a subset of AI, deriving meaningful information from visual input such as images and videos and, in some cases, acting based on its interpretations [4]. With CV having gained substantial attention in numerous industries, including automotive, health care, energy, and manufacturing [5], the expectation that this technology might abet disruptive innovations in educational and learning contexts is growing. A. Examples of Computer Vision in Education Research CV has been examined across a range of educational and learning contexts. In one example, researchers examined student engagement within classroom recordings [6]. Findings suggest that CV can be used to recognize individual behaviors with 'reasonable accuracy' and as a way of quantifying in-class behaviors. CV is also being widely considered to detect and assess live learner and instructor interactions as a way of providing actionable feedback [7]. Researchers have examined how learners attend to objects and interact with instructional videos [8] and how CV can aid in the improvement and development of instructional materials [9]. However, the majority of extant CV and education research has been applied with neurotypical learners. Research exploring how CV methods and processes might be applied for neurodiverse learners is limited, despite evidence that this approach has vast potential for supporting neurodiverse groups [10]. In addition to this, most of the research tends to follow what is known as a medical model, which focuses on early detection and diagnosis of neurodiverse characteristics [10]. For example, CV methods have been deployed to analyze attention and psychological factors encoded in eye movements and to develop emotional classifiers [11] of individuals to help in diagnosis of autism [12]. However, a more social-ecological perspective for incorporating CV might consider how researchers could leverage this technology to better understand how learners attend to and engage in technology-mediated learning environments and multimedia [13]. Whereas most prior research in this area has utilized eye tracking to make inferences about how neurodiverse learners attend to objects of importance [15], advancements in CV provide new opportunities to examine this area. II. VIRTUOSO-SVVR The research presented within this work-in-progress paper reports the approach one team of researchers has developed to implement CV as a way of examining usage patterns of neurodiverse learners within a 360-degree spherical video-based virtual reality (SVVR) system called Virtuoso-SVVR [16]. Authorized licensed use limited to: UNIVERSITY OF CONNECTICUT. Downloaded on July 13,2022 at 17:32:27 UTC from IEEE Xplore. Restrictions apply.

There are theoretical arguments and empirical studies to support the benefits of early participation in an online course. However, little research has been conducted to investigate the relationship between students’ early start in online learning and their course performance, which limits its application to online education research and practice. This study aims to fill this gap by examining and quantifying the relationship of early participation in course activities with students’ learning performance. Using multilevel logit modeling, the relationship of early participation in online courses and student final performance are modeled with student log data and other individual information from over 30,000 students enrolled in 22 online university courses. Results show that early participation in online course activities is significantly correlated with student final performance. As the first study to quantify the relationship between early participation and students’ academic performance, this paper provides important insights for online teaching and learning research.

Socially shared metacognition is important for effective collaborative problem solving in virtual laboratory settings, A holistic account of socially shared metacognition in virtual laboratory settings is needed to advance our understanding, but previous studies have only focused on the isolated effect of each dimension on problem solving. This study thus applied learning analytics techniques to develop a comprehensive understanding of socially shared metacognition during collaborative problem solving in virtual laboratories. We manually coded 126 collaborative problem-solving scenarios in a virtual physics laboratory and then employed K-Means clustering analysis to identify patterns of socially shared metacognition. Four clusters were discovered. Statistical analysis was performed to investigate how the clusters were associated with the outcome of collaborative problem solving and also how they related to the difficulty level of problems. The findings of this study provided theoretical implications to advance the understanding of socially shared metacognition in virtual laboratory settings and also practical implications to foster effective collaborative problem solving in those settings.

This study explored youth with Autism Spectrum Disorder (ASD) learning social competence in the context of innovative 3D virtual learning environment and the effects of gaming as a central element of the learning experience. The empirical study retrospectively compared the social interactions of 11 adolescents with ASD in game-and nongame-based 3D collaborative learning activities in the same social competence training curriculum. We employed a learning analytics approach - association rule mining to uncover the associative rules of verbal social interaction and nonverbal social interaction contributors from the large dataset of the coded social behaviors. By comparing the rules across the game and nongame activities, we found a significant difference in youth with ASD’s social performance. The results of the group comparison study indicated that the co-occurrence of verbal and nonverbal behaviors is much stronger in the game-based learning activities. The game activities also yielded more diverse social interaction behavior patterns. On the other hand, in the nongame activities, students’ social interaction behavior patterns are much more limited. Furthermore, the impact of game design principles on learning is then discussed in this paper.

With the advancement of digital technologies, big data and learning analytics have become prevalent in the higher education. Various student-facing systems increased the amount of data available to students, and whether students can use big data and learning analytics effectively will affect their academic success. Most studies, however, have focused on how teachers and administrative personnel use student data to make data-driven instruction and management decisions. As a result, little attention has been given to students' use of relevant data that generated by big data and learning analytics to promote their own learning and growth. This study explored using social cognitive theory to identify possible environmental, personal, and behavioural factors that affect students' data use. We used an online questionnaire that collected 242 completed surveys from Chinese university students. Partial Least Squares (PLS) path modelling was used to analyse the data. The initial findings support the conclusion that university students could be encouraged to effectively use data in three ways: (1) through the promotion of university-wide cultures of data use and sustained improvements in data quality, (2) through the professional development of student data literacy, and (3) through the support of student data autonomy, student data reflectiveness, and students' digital identities.