🤖 AI Summary
This work addresses a key limitation in traditional IMU-based movement assessment methods, which rely on one-hot labels and thereby ignore the inherent ambiguity in class boundaries across repeated actions—failing to capture the reasonable disagreement often observed among human raters. To overcome this, the authors propose a method that automatically generates label distributions without requiring extensive manual scoring, leveraging KL divergence to jointly optimize a deep network for both high-accuracy classification and detection of ambiguous samples. Evaluated on four IMU datasets, the approach matches or surpasses cross-entropy baselines in classification performance while reliably identifying ambiguous instances and their associated classes. This represents the first effective modeling and utilization of label ambiguity in IMU-based movement evaluation.
📝 Abstract
Home-based physiotherapy is performed without supervision, which leads to incorrect execution and motivates systems that assess movement automatically from inertial measurement units (IMUs). Such systems assign each repetition to a category, yet a relevant share of repetitions falls near a class boundary, where even trained raters disagree. Classifiers trained with one-hot labels collapse these borderline repetitions onto a single class and discard this ambiguity. We address this with a method that automatically generates a label distribution per repetition without a large rater pool. We train a network to reproduce the full distribution with a Kullback-Leibler objective, the ambiguity approach, and compare it against a one-hot cross-entropy baseline on four IMU exercise datasets. From the network output we further determine whether a repetition is ambiguous and which classes are relevant to it. The ambiguity approach matched or exceeded the baseline classification on all four datasets, and detected ambiguity and the relevant classes more reliably. Representing the label distribution in the training target therefore adds information about ambiguity at no cost to classification.