Toward Automated Classroom Observation: Multimodal Machine Learning to Estimate CLASS Positive Climate and Negative Climate
نویسندگان
چکیده
In this work we present a multi-modal machine learning-based system, which call ACORN, to analyze videos of school classrooms for the Positive Climate (PC) and Negative (NC) dimensions CLASS observation protocol that is widely used in educational research. ACORN uses convolutional neural networks spectral audio features, faces teachers students, pixels each image frame, then integrates information over time using Temporal Convolutional Networks. The audiovisual ACORN's PC NC predictions have Pearson correlations $0.55$ $0.63$ with ground-truth scores provided by expert coders on UVA Toddler dataset (cross-validation $n=300$ 15-min video segments), purely auditory predicts $0.36$ $0.41$ MET (test set $n=2000$ segments). These numbers are similar inter-coder reliability human coders. Finally, Graph Networks make early strides (AUC=$0.70$) toward predicting specific moments (45-90sec clips) when particularly weak/strong. Our findings inform design automatic classroom also more general activity recognition summary systems.
منابع مشابه
Nurses' attitude toward supportive work climate affecting transfer of learning to job
Introduction: Transfer of training has been defined as the application of new knowledge, skills, and attitudes learned from continuing education programs to the job. Learning transfer can be influenced by many factors that can facilitate or hinder it. A supportive work climate is crucial for successful transfer of learning to job. The aim of this study was to evaluate the nurses' attitude towar...
متن کاملTeacher Personality and Classroom Climate
DESCRIPTORSBIBLIOGRAPHIES, *INTERACTION, LITERATURE REVIEWS, MALES, *PERSONALITY STUDIES, *PERSONALITY THEORIES, PHYSICS TEACHERS, QUESTIONNAIRES, SECONDARY SCHOOL TEACHERS, *STUDENT CHARACTERISTICS, TABLES (DATA), TEACHER ATTITUDES, *TEACHER CHARACTERISTICS, MINNESOTA TEACHER ATTITUDE INVENTORY (MTAI), CLASSROOM CLIMATE QUESTIONNAIRE (CCQ), EDWARDS PERSONAL PREFERENCE SCHEDULE (EPPS), ALLPORT ...
متن کاملEffects of Game Tournaments on Learning and Classroom Climate
In this paper, we motivate and describe a term long game tournament. The tournament has different milestones, including breadth first search, heuristic search, knowledge representation and reasoning and the tournament proper. For this last milestone, students are encouraged to use machine learning techniques. We describe in detail each of the milestones as well as the game and its specification...
متن کاملEvaluating machine learning methods and satellite images to estimate combined climatic indices
The reflections recorded on satellite images have been affected by various environmental factors. In these images, some of these factors are combined with other environmental factors that cannot be distinguished. Therefore, it seems wise to model these environmental phenomena in the form of hybrid indicators. In this regard, satellite imagery and machine learning methods can play a unique role ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE Transactions on Affective Computing
سال: 2023
ISSN: ['1949-3045', '2371-9850']
DOI: https://doi.org/10.1109/taffc.2021.3059209