NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·2h ago
AI Research
arXiv CS.AI·May 25·5 min read
An Interpretable Closed-Loop Intelligent Tutoring System for Multimodal Affective Feedback in Asynchronous Presentation Training
TL;DR
Many learners struggle to receive effective feedback on their presentation skills, which can hinder their development. An interpretable closed-loop Intelligent Tutoring System (ITS) was developed, utilizing a seven-dimensional Behaviorally Anchored Rating Scale (BARS) and multimodal inputs to provide personalized feedback.
✦ Why It Matters
Engineers can leverage multimodal feedback systems to enhance learning outcomes in educational technologies.
Key Takeaways
- Implement multimodal feedback for personalized learning experiences.
- Utilize XGBoost for effective performance assessment in educational tools.
- Incorporate structured feedback mechanisms to drive skill improvement.
- Explore the use of BARS for evaluating complex competencies.
Full Summary
Effective feedback is crucial for developing presentation skills, yet traditional methods often lack personalization and scalability. An interpretable closed-loop Intelligent Tutoring System (ITS) was created, which employs a seven-dimensional Behaviorally Anchored Rating Scale (BARS) to assess learners' performance.
The system integrates multimodal inputs, including facial expressions, vocal tone, text, and eye movement, to generate evidence-based feedback. Built on an XGBoost model, it was trained using 10,360 video segments from Massive Open Online Courses (MOOCs) and achieved scoring accuracy comparable to expert ratings.
In a study with 204 adult learners over 30 days, significant improvements were observed in all BARS dimensions, with practice frequency positively correlating with performance. These findings highlight the potential of multimodal analytics in driving behavioral change through structured feedback, paving the way for more effective Intelligent Tutoring Systems in education.