Interaction Datasets

Interaction datasets are collections of annotated data capturing human-human, human-object, or even multi-party interactions, primarily used to train and evaluate machine learning models for behavior analysis. Current research focuses on developing more complex datasets representing nuanced interactions (e.g., loose collaborations, rapport in children), employing diverse model architectures like graph neural networks, LSTMs, and convolutional neural networks to analyze these data, and addressing challenges such as data bias and interpretability. These datasets and associated models have significant implications for various fields, including healthcare (e.g., autism diagnosis), robotics (e.g., human-robot interaction), and education (e.g., understanding student collaboration).