We address the challenging task of identifying, segmenting, and tracking hand-held objects, which is crucial for applications such as human action segmentation and per-formance evaluation. This task is particularly challenging due to heavy occlusion, rapid motion, and the transitory nature of objects being hand-held, where an object may be held, released, and subsequently picked up again. To tackle these challenges, we have developed a novel transformer-based architecture called HOIST-Former. HOIST-Former is adept at spatially and temporally segmenting hands and objects by iteratively pooling features from each other, ensuring that the processes of identification, segmentation, and tracking of hand-held objects depend on the hands' positions and their contextual appearance. We further refine HOIST-Former with a contact loss that focuses on areas where hands are in contact with objects. Moreover, we also contribute an in-the-wild video dataset called HOIST, which comprises 4,125 videos complete with bounding boxes, segmentation masks, and tracking IDs for hand-held objects. Through experiments on the HOIST dataset and two ad-ditional public datasets, we demonstrate the efficacy of HOIST-Former in segmenting and tracking hand-held objects. Project page: https://supreethn.github.io/research/hoistformer/index.html

Multiple-choice questions, which are widely used in educational assessments, have the potential to negatively impact student learning and skew analytics when they contain item-writing flaws. Existing methods for evaluating multiple-choice questions in educational contexts tend to focus primarily on machine readability metrics, such as grammar, syntax, and formatting, without considering the intended use of the questions within course materials and their pedagogical implications. In this study, we present the results of crowdsourcing the evaluation of multiple-choice questions based on 15 common item-writing flaws. Through analysis of 80 crowdsourced evaluations on questions from the domains of calculus and chemistry, we found that crowdworkers were able to accurately evaluate the questions, matching 75% of the expert evaluations across multiple questions. They were able to correctly distinguish between two levels of Bloom's Taxonomy for the calculus questions, but were less accurate for chemistry questions. We discuss how to scale this question evaluation process and the implications it has across other domains. This work demonstrates how crowdworkers can be leveraged in the quality evaluation of educational questions, regardless of prior experience or domain knowledge.

Scientists and journalists strive to report numbers with high precision to keep readers well-informed. Our work investigates whether this practice can backfire due to the cognitive costs of processing multi-digit precise numbers. In a pre-registered randomized experiment, we presented readers with several news stories containing numbers in either precise or round versions. We then measured their ability to approximately recall these numbers and make estimates based on what they read. Our results revealed a counter-intuitive effect where reading round numbers helped people better approximate the precise values, while seeing precise numbers made them worse. We also conducted two surveys to elicit individual preferences for the ideal degree of rounding for numbers spanning seven orders of magnitude in various contexts. From the surveys, we found that people tended to prefer more precision when the rounding options contained only digits (e.g., ”2,500,000”) than when they contained modifier terms (e.g., ”2.5 million”). We conclude with a discussion of how these findings can be leveraged to enhance numeracy in digital content consumption.

Starting from the spring of 2020, higher institutions in the US underwent a rapid shift from in-person classes to emergency remote education, in response to the COVID-19 outbreak. Under this circumstance, a variety of video conferencing tools (e.g., Zoom) have been adopted for distance education, which pose a set of new challenges arising from synchronous online classes. Among these, one significant issue was students' unwillingness to open cameras, resulting in a lack of non-verbal cues that instructors could rely on to gauge students' understanding and adjust their teachings. Towards addressing this issue, our qualitative study aims at investigating the rationales behind students' camera avoidance. Through a series of semi-structured interviews on undergraduate students in the U.S, we identified prominent factors -- namely the class size, lecture style, level of interactivity and privacy concerns -- that influenced students' motivation for opening their cameras. At the same time, we uncovered several difficulties, such as heightened self-awareness, feeling of minority and academic perspective, that discouraged students from opening camera, with more substantial impacts on international students. We conclude with actionable insights into the design of online classes, video-conferencing platforms and camera technology that can promote camera usage, thereby contributing to scalable and inclusive interventions for facilitating the transition into remote education.

While data science education has gained increased recognition in both academic institutions and industry, there has been a lack of research on automated coding assessment for novice students. Our work presents a first step in this direction, by leveraging the coding metrics from traditional software engineering (Halstead Volume and Cyclomatic Complexity) in combination with those that reflect a data science project’s learning objectives (number of library calls and number of common library calls with the solution code). Through these metrics, we examined the code submissions of 97 students across two semesters of an introductory data science course. Our results indicated that the metrics can identify cases where students had overly complicated codes and would benefit from scaffolding feedback. The number of library calls, in particular, was also a significant predictor of changes in submission score and submission runtime, which highlights the distinctive nature of data science programming. We conclude with suggestions for extending our analyses towards more actionable intervention strategies, for example by tracking the fine-grained submission grading outputs throughout a student’s submission history, to better model and support them in their data science learning process.

While generating multiple-choice questions has been shown to promote deep learning, students often fail to realize this benefit and do not willingly participate in this activity. Additionally, the quality of the student-generated questions may be influenced by both their level of engagement and familiarity with the learning materials. Towards better understanding how students can generate high quality questions, we designed and deployed a multiple-choice question generation activity in seven college-level online chemistry courses. From these courses, we collected data on student interactions and their contribution to the question-generation task. A total of 201 students enrolled in the courses and 57 of them elected to generate a multiple-choice question. Our results indicated that students were able to contribute quality questions, with 67% of them being evaluated by experts as acceptable for use. We further identified several student behaviors in the online courses that are correlated to their participation in the task and the quality of their contribution. Our findings can help teachers and students better understand the benefits of student-generated questions and effectively implement future learnersourcing activities.

Virtual-Reality (VR), Augmented-Reality (AR) and Mixed-Reality (MR) are finding new applications in clinical investigations. This study uses VR, AR, and MR techniques for pre-surgery applications on three platforms: Windows PC, Apple iPad, and Mobile. Using the proposed technique, which involves the use of virtual reality, augmented reality, and mixed reality technologies, a physician can visualize the problem and discuss options with a patient while the patient's anatomy is being studied. The combination of endoscopic visualization can be used for surgical planning, training, medical diagnosis, and postoperative examination.

Assigning a set of hypothesized knowledge components (KCs) to assessment items within an ed-tech system enables us to better estimate student learning. However, creating and assigning these KCs is a time-consuming process that often requires domain expertise. In this study, we present the results of crowdsourcing KCs for problems in the domain of mathematics and English writing, as a first step in leveraging the crowd to expedite this task. Crowdworkers were presented with a problem and asked to provide the underlying skills required to solve it. Additionally, we investigated the effect of priming crowdworkers with related content before having them generate these KCs. We then analyzed their contributions through qualitative coding and found that across both the math and writing domains roughly 33% of the crowdsourced KCs directly matched those generated by domain experts for the same problems.

How do the features of a learning environment's user interface impact learners' agency and, further, their learning? We explored this question in the context of Decimal Point, a digital learning game designed to support middle school students in learning decimals. Previous studies of the game showed that giving students the ability to choose the order and number of mini-games to play did not significantly impact their learning outcomes compared to a condition without choice. In this paper we explore whether some elements of the game's interface may have inadvertently exerted indirect control over students' choice, leading to the previous effects. We conducted a classroom study using a new version of the game that varied whether students saw a visual path connecting mini-games on the game map to modulate the level of indirect control students would experience with an implied ordering. Ultimately, we found that students in the no-line condition exercised significantly more agency but did not learn any less than the line condition. These results suggest that indirect control can be a subtle but powerful way to direct student attention in digital learning games.

Passive monitoring utilizing distributed wireless sniffers is an effective technique to monitor activities in wireless infrastructure networks for fault diagnosis, resource management and critical path analysis. In this paper, we introduce a quality of monitoring (QoM) metric defined by the expected number of active users monitored, and investigate the problem of maximizing QoM by judiciously assigning sniffers to channels based on knowledge of user activities in a multi-channel wireless network. Two capture models are considered. The first one, called the user-centric model assumes frame-level capturing capability of sniffers such that the activities of different users can be distinguished. The second one, called the sniffer-centric model only utilizes binary channel information (active or not) at a sniffer. For the user-centric model, we show that the implied optimization problem is NP-hard, but a constant approximation ratio can be attained via polynomial complexity algorithms. For the sniffer-centric model, we devise a stochastic inference scheme that transforms the problem into the user-centric domain, where we are able to apply our polynomial approximation algorithms. The effectiveness of our proposed scheme and algorithms is further evaluated using both synthetic data as well as real-world traces from an operational WLAN.

Bandwidth allocation is a fundamental problem in communication networks. With current network moving towards the future Internet model, the problem is further intensified as network traffic demanding far from exceeds network bandwidth capability. Maintaining a certain user satisfaction degree therefore becomes a challenge research topic. In this paper, we deal with the problem by proposing BASMIN, a novel bandwidth allocation scheme that aims to maximize network userpsilas happiness. We also defined a new metric for evaluating network user satisfaction degree: network worth. A three-step evaluation process is then conducted to compare BASMIN efficiency with other three popular bandwidth allocation schemes. Throughout the tests, we experienced BASMIN's advantages over the others; we even found out that one of the most widely used bandwidth allocation schemes, in fact, is not effective at all.