Confidence Estimation via Wrist Movement
Seiya Tanaka, Andrew W. Vargo, Motoi Iwata, Koichi Kise · 2021
Recent advancements have shown that physiological sensing techniques such as eye-gaze and handwriting can be used to classify user confidence when answering questions. While accurately classifying confidence is important for providing feedback to learners, simplifying and streamlining the process is important for real-world applications. The aim of this study is to use a single accelerometer on the wrist to estimate confidence, thus reducing the burden of device orientation and cost for users. We present a study of 10 participants who wrote answers while wearing a wrist accelerometer. Using SVM, we achieved an accuracy rate of 83.8% for participant-dependent and 77.0% for participant-independent classification when the answer given by the user was correct and 84.4% and 71.9% when the answer was incorrect. The results indicate that using a simple wrist accelerometer is effective for providing confidence estimation.