Agents must monitor their partners' affective states continuously in order to understand and engage in social interactions. However, methods for evaluating affect recognition do not account for changes in classification performance that may occur during occlusions or transitions between affective states. This paper addresses temporal patterns in affect classification performance in the context of an infant-robot interaction, where infants’ affective states contribute to their ability to participate in a therapeutic leg movement activity. To support robustness to facial occlusions in video recordings, we trained infant affect recognition classifiers using both facial and body features. Next, we conducted an in-depth analysis of our best-performing models to evaluate how performance changed over time as the models encountered missing data and changing infant affect. During time windows when features were extracted with high confidence, a unimodal model trained on facial features achieved the same optimal performance as multimodal models trained on both facial and body features. However, multimodal models outperformed unimodal models when evaluated on the entire dataset. Additionally, model performance was weakest when predicting an affective state transition and improved after multiple predictions of the same affective state. These findings emphasize the benefits of incorporating body features in continuous affect recognition for infants. Our work highlights the importance of evaluating variability in model performance both over time and in the presence of missing data when applying affect recognition to social interactions.
more »
« less
Multimodal Engagement Analysis from Facial Videos in the Classroom
Student engagement is a key component of learning and teaching, resulting in a plethora of automated methods to measure it. Whereas most of the literature explores student engagement analysis using computer-based learning often in the lab, we focus on using classroom instruction in authentic learning environments. We collected audiovisual recordings of secondary school classes over a one and a half month period, acquired continuous engagement labeling per student (N=15) in repeated sessions, and explored computer vision methods to classify engagement from facial videos. We learned deep embeddings for attentional and affective features by training Attention-Net for head pose estimation and Affect-Net for facial expression recognition using previously-collected large-scale datasets. We used these representations to train engagement classifiers on our data, in individual and multiple channel settings, considering temporal dependencies. The best performing engagement classifiers achieved student-independent AUCs of .620 and .720 for grades 8 and 12, respectively, with attention-based features outperforming affective features. Score-level fusion either improved the engagement classifiers or was on par with the best performing modality. We also investigated the effect of personalization and found that only 60 seconds of person-specific data, selected by margin uncertainty of the base classifier, yielded an average AUC improvement of .084.
more »
« less
- NSF-PAR ID:
- 10349953
- Date Published:
- Journal Name:
- IEEE Transactions on Affective Computing
- ISSN:
- 2371-9850
- Page Range / eLocation ID:
- 1 to 1
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
The overall goal of our research is to develop a system of intelligent multimodal affective pedagogical agents that are effective for different types of learners (Adamo et al., 2021). While most of the research on pedagogical agents tends to focus on the cognitive aspects of online learning and instruction, this project explores the less-studied role of affective (or emotional) factors. We aim to design believable animated agents that can convey realistic, natural emotions through speech, facial expressions, and body gestures and that can react to the students’ detected emotional states with emotional intelligence. Within the context of this goal, the specific objective of the work reported in the paper was to examine the extent to which the agents’ facial micro-expressions affect students’ perception of the agents’ emotions and their naturalness. Micro-expressions are very brief facial expressions that occur when a person either deliberately or unconsciously conceals an emotion being felt (Ekman &Friesen, 1969). Our assumption is that if the animated agents display facial micro expressions in addition to macro expressions, they will convey higher expressive richness and naturalness to the viewer, as “the agents can possess two emotional streams, one based on interaction with the viewer and the other based on their own internal state, or situation” (Queiroz et al. 2014, p.2).The work reported in the paper involved two studies with human subjects. The objectives of the first study were to examine whether people can recognize micro-expressions (in isolation) in animated agents, and whether there are differences in recognition based on the agent’s visual style (e.g., stylized versus realistic). The objectives of the second study were to investigate whether people can recognize the animated agents’ micro-expressions when integrated with macro-expressions, the extent to which the presence of micro + macro-expressions affect the perceived expressivity and naturalness of the animated agents, the extent to which exaggerating the micro expressions, e.g. increasing the amplitude of the animated facial displacements affects emotion recognition and perceived agent naturalness and emotional expressivity, and whether there are differences based on the agent’s design characteristics. In the first study, 15 participants watched eight micro-expression animations representing four different emotions (happy, sad, fear, surprised). Four animations featured a stylized agent and four a realistic agent. For each animation, subjects were asked to identify the agent’s emotion conveyed by the micro-expression. In the second study, 234 participants watched three sets of eight animation clips (24 clips in total, 12 clips per agent). Four animations for each agent featured the character performing macro-expressions only, four animations for each agent featured the character performing macro- + micro-expressions without exaggeration, and four animations for each agent featured the agent performing macro + micro-expressions with exaggeration. Participants were asked to recognize the true emotion of the agent and rate the emotional expressivity ad naturalness of the agent in each clip using a 5-point Likert scale. We have collected all the data and completed the statistical analysis. Findings and discussion, implications for research and practice, and suggestions for future work will be reported in the full paper. ReferencesAdamo N., Benes, B., Mayer, R., Lei, X., Meyer, Z., &Lawson, A. (2021). Multimodal Affective Pedagogical Agents for Different Types of Learners. In: Russo D., Ahram T., Karwowski W., Di Bucchianico G., Taiar R. (eds) Intelligent Human Systems Integration 2021. IHSI 2021. Advances in Intelligent Systems and Computing, 1322. Springer, Cham. https://doi.org/10.1007/978-3-030-68017-6_33Ekman, P., &Friesen, W. V. (1969, February). Nonverbal leakage and clues to deception. Psychiatry, 32(1), 88–106. https://doi.org/10.1080/00332747.1969.11023575 Queiroz, R. B., Musse, S. R., &Badler, N. I. (2014). Investigating Macroexpressions and Microexpressions in Computer Graphics Animated Faces. Presence, 23(2), 191-208. http://dx.doi.org/10.1162/
-
null (Ed.)Over the past two decades, educators have used computer-supported collaborative learning (CSCL) to integrate technology with pedagogy to improve student engagement and learning outcomes. Researchers have also explored the diverse affordances of CSCL, its contributions to engineering instruction, and its effectiveness in K-12 STEM education. However, the question of how students use CSCL resources in undergraduate engineering classrooms remains largely unexplored. This study examines the affordances of a CSCL environment utilized in a sophomore dynamics course with particular attention given to the undergraduate engineering students’ use of various CSCL resources. The resources include a course lecturebook, instructor office hours, a teaching assistant help room, online discussion board, peer collaboration, and demonstration videos. This qualitative study uses semi-structured interview data collected from nine mechanical engineering students (four women and five men) who were enrolled in a dynamics course at a large public research university in Eastern Canada. The interviews focused on the individual student’s perceptions of the school, faculty, students, engineering courses, and implemented CSCL learning environment. The thematic analysis was conducted to analyze the transcribed interviews using a qualitative data analysis software (Nvivo). The analysis followed a six step process: (1) reading interview transcripts multiple times and preliminary in vivo codes; (2) conducting open coding by coding interesting or salient features of the data; (3) collecting codes and searching for themes; (4) reviewing themes and creating a thematic map; (5) finalizing themes and their definitions; and (6) compiling findings. This study found that the students’ use of CSCL resources varied depending on the students’ personal preferences, as well as their perceptions of the given resource’s value and its potential to enhance their learning. For example, the dynamics lecturebook, which had been redesigned to encourage problem solving and note-taking, fostered student collaborative problem solving with their peers. In contrast, the professor’s example video solutions had much more of an influence on students’ independent problem-solving processes. The least frequently used resource was the course’s online discussion forum, which could be used as a means of communication. The findings reveal how computer-supported collaborative learning (CSCL) environments enable engineering students to engage in multiple learning opportunities with diverse and flexible resources to both address and to clarify their personal learning needs. This study strongly recommends engineering instructors adapt a CSCL environment for implementation in their own unique classroom context.more » « less
-
Michalsky, Tova ; Moos, Daniel (Ed.)Teachers’ ability to self-regulate their own learning is closely related to their competency to enhance self-regulated learning (SRL) in their students. Accordingly, there is emerging research for the design of teacher dashboards that empower instructors by providing access to quantifiable evidence of student performance and SRL processes. Typically, they capture evidence of student learning and performance to be visualized through activity traces (e.g., bar charts showing correct and incorrect response rates, etc.) and SRL data (e.g., eye-tracking on content, log files capturing feature selection, etc.) in order to provide teachers with monitoring and instructional tools. Critics of the current research on dashboards used in conjunction with advanced learning technologies (ALTs) such as simulations, intelligent tutoring systems, and serious games, argue that the state of the field is immature and has 1) focused only on exploratory or proof-of-concept projects, 2) investigated data visualizations of performance metrics or simplistic learning behaviors, and 3) neglected most theoretical aspects of SRL including teachers’ general lack of understanding their’s students’ SRL. Additionally, the work is mostly anecdotal, lacks methodological rigor, and does not collect critical process data (e.g. frequency, duration, timing, or fluctuations of cognitive, affective, metacognitive, and motivational (CAMM) SRL processes) during learning with ALTs used in the classroom. No known research in the areas of learning analytics, teacher dashboards, or teachers’ perceptions of students’ SRL and CAMM engagement has systematically and simultaneously examined the deployment, temporal unfolding, regulation, and impact of all these key processes during complex learning. In this manuscript, we 1) review the current state of ALTs designed using SRL theoretical frameworks and the current state of teacher dashboard design and research, 2) report the important design features and elements within intelligent dashboards that provide teachers with real-time data visualizations of their students’ SRL processes and engagement while using ALTs in classrooms, as revealed from the analysis of surveys and focus groups with teachers, and 3) propose a conceptual system design for integrating reinforcement learning into a teacher dashboard to help guide the utilization of multimodal data collected on students’ and teachers’ CAMM SRL processes during complex learning.more » « less
-
Since the 2014 high-profile meta-analysis of undergraduate STEM courses, active learning has become a standard in higher education pedagogy. One way to provide active learning is through the flipped classroom. However, finding suitable pre-class learning activities to improve student preparation and the subsequent classroom environment, including student engagement, can present a challenge in the flipped modality. To address this challenge, adaptive learning lessons were developed for pre-class learning for a course in Numerical Methods. The lessons would then be used as part of a study to determine their cognitive and affective impacts. Before the study could be started, it involved constructing well-thought-out adaptive lessons. This paper discusses developing, refining, and revising the adaptive learning platform (ALP) lessons for pre-class learning in a Numerical Methods flipped course. In a prior pilot study at a large public southeastern university, the first author had developed ALP lessons for the pre-class learning for four (Nonlinear Equations, Matrix Algebra, Regression, Integration) of the eight topics covered in a Numerical Methods course. In the current follow-on study, the first author and two other instructors who teach Numerical Methods, one from a large southwestern urban university and another from an HBCU, collaborated on developing the adaptive lessons for the whole course. The work began in Fall 2020 by enumerating the various chapters and breaking each one into individual lessons. Each lesson would include five sections (introduction, learning objectives, video lectures, textbook content, assessment). The three instructors met semi-monthly to discuss the content that would form each lesson. The main discussion of the meetings centered on what a student would be expected to learn before coming to class, choosing appropriate content, agreeing on prerequisites, and choosing and making new assessment questions. Lessons were then created by the first author and his student team using a commercially available platform called RealizeIT. The content was tested by learning assistants and instructors. It is important to note that significant, if not all, parts of the content, such as videos and textbook material, were available through previously done work. The new adaptive lessons and the revised existing ones were completed in December 2020. The adaptive lessons were tested for implementation in Spring 2021 at the first author's university and made 15% of the students' grade calculation. Questions asked by students during office hours, on the LMS discussion board, and via emails while doing the lessons were used to update content, clarify questions, and revise hints offered by the platform. For example, all videos in the ALP lessons were updated to HD quality based on student feedback. In addition, comments from the end-of-semester surveys conducted by an independent assessment analyst were collated to revise the adaptive lessons further. Examples include changing the textbook content format from an embedded PDF file to HTML to improve quality and meet web accessibility standards. The paper walks the reader through the content of a typical lesson. It also shows the type of data collected by the adaptive learning platform via three examples of student interactions with a single lesson.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1109/TAFFC.2021.3127692
