Search for: All records

Large language models’ (LLMs) abilities are drawn from their pretraining data, and model development begins with data curation. However, decisions around what data is retained or removed during this initial stage are underscrutinized. In our work, we ground web text, which is a popular pretraining data source, to its social and geographic contexts. We create a new dataset of 10.3 million self-descriptions of website creators, and extract information about who they are and where they are from: their topical interests, social roles, and geographic affiliations. Then, we conduct the first study investigating how ten “quality” and English language identification (langID) filters affect webpages that vary along these social dimensions. Our experiments illuminate a range of implicit preferences in data curation: we show that some quality classifiers act like topical domain filters, and langID can overlook English content from some regions of the world. Overall, we hope that our work will encourage a new line of research on pretraining data curation practices and its social implications. 

Agents must monitor their partners' affective states continuously in order to understand and engage in social interactions. However, methods for evaluating affect recognition do not account for changes in classification performance that may occur during occlusions or transitions between affective states. This paper addresses temporal patterns in affect classification performance in the context of an infant-robot interaction, where infants’ affective states contribute to their ability to participate in a therapeutic leg movement activity. To support robustness to facial occlusions in video recordings, we trained infant affect recognition classifiers using both facial and body features. Next, we conducted an in-depth analysis of our best-performing models to evaluate how performance changed over time as the models encountered missing data and changing infant affect. During time windows when features were extracted with high confidence, a unimodal model trained on facial features achieved the same optimal performance as multimodal models trained on both facial and body features. However, multimodal models outperformed unimodal models when evaluated on the entire dataset. Additionally, model performance was weakest when predicting an affective state transition and improved after multiple predictions of the same affective state. These findings emphasize the benefits of incorporating body features in continuous affect recognition for infants. Our work highlights the importance of evaluating variability in model performance both over time and in the presence of missing data when applying affect recognition to social interactions. 

Observing how infants and mothers coordinate their behaviors can highlight meaningful patterns in early communication and infant development. While dyads often differ in the modalities they use to communicate, especially in the first year of life, it remains unclear how to capture coordination across multiple types of behaviors using existing computational models of interpersonal synchrony. This paper explores Dynamic Mode Decomposition with control (DMDc) as a method of integrating multiple signals from each communicating partner into a model of multimodal behavioral coordination. We used an existing video dataset to track the head pose, arm pose, and vocal fundamental frequency of infants and mothers during the Face-to-Face Still-Face (FFSF) procedure, a validated 3-stage interaction paradigm. For each recorded interaction, we fit both unimodal and multimodal DMDc models to the extracted pose data. The resulting dynamic characteristics of the models were analyzed to evaluate trends in individual behaviors and dyadic processes across infant age and stages of the interactions. Results demonstrate that observed trends in interaction dynamics across stages of the FFSF protocol were stronger and more significant when models incorporated both head and arm pose data, rather than a single behavior modality. Model output showed significant trends across age, identifying changes in infant movement and in the relationship between infant and mother behaviors. Models that included mothers’ audio data demonstrated similar results to those evaluated with pose data, confirming that DMDc can leverage different sets of behavioral signals from each interacting partner. Taken together, our results demonstrate the potential of DMDc toward integrating multiple behavioral signals into the measurement of multimodal interpersonal coordination. 

Early intervention to address developmental disability in infants has the potential to promote improved outcomes in neurodevelopmental structure and function [1]. Researchers are starting to explore Socially Assistive Robotics (SAR) as a tool for delivering early interventions that are synergistic with and enhance human-administered therapy. For SAR to be effective, the robot must be able to consistently attract the attention of the infant in order to engage the infant in a desired activity. This work presents the analysis of eye gaze tracking data from five 6-8 month old infants interacting with a Nao robot that kicked its leg as a contingent reward for infant leg movement. We evaluate a Bayesian model of lowlevel surprise on video data from the infants’ head-mounted camera and on the timing of robot behaviors as a predictor of infant visual attention. The results demonstrate that over 67% of infant gaze locations were in areas the model evaluated to be more surprising than average. We also present an initial exploration using surprise to predict the extent to which the robot attracts infant visual attention during specific intervals in the study. This work is the first to validate the surprise model on infants; our results indicate the potential for using surprise to inform robot behaviors that attract infant attention during SAR interactions.