This study proposes the use of multimodal fusion models employing Convolutional Neural Networks (CNNs) to extract meeting minutes from group discussion corpus. First, unimodal models are created using raw behavioral data such as speech, head motion, and face tracking. These models are then integrated into a fusion model that works as a classifier. The main advantage of this work is that the proposed models were trained without any hand-crafted features, and they outperformed a baseline model that was trained using hand-crafted features. It was also found that multimodal fusion is useful in applying the CNN approach to model multimodal multiparty interaction.

Gesture generation is one of the most important tasks in humanoid interfaces because hand gestures by humanoid robots and animated agents are useful in improving the comprehensibility of conversation content. This study proposes a method for automatically generating iconic drawing gestures using image processing and machine learning techniques. First, we collected a set of graphic images for over 1000 objects and classified the objects into 4 types of shapes; these shapes were used as the drawing gesture shapes. By implementing a gesture shape decision mechanism, we also built a system that takes a sentence as the system input and produces hand gesture animations that are synchronized with synthetic speech.

In order to provide a watching service for the elderly with dementia, recognizing and assessing their cognitive and health status are indispensable. In this study, we propose a prediction model for assessing the communication attitude of the elderly during interacting with a virtual agent. We define speech features and head motion features using frequency analysis and apply them to a linear regression analysis. The coefficient of determination for the model using only speech features was 0.413 and that for a model that exploited both speech and head movement features was 0.505. This result suggests that combining speech and head movement data is useful in predicting the communication attitude of the elderly, and the model is can be applied for automatic assessment in watching services.

Group discussions are used for various purposes, such as creating new ideas and making a group decision. It is desirable to archive the results and processes of the discussion as useful resources for the group. Therefore, a key technology would be a way to extract meaningful resources from a group discussion. To accomplish this goal, we propose classification models that select meeting extracts to be included in the discussion summary based on nonverbal behavior such as attention, head motion, prosodic features, and co-occurrence patterns of these behaviors. We create different prediction models depending on the degree of extract-worthiness, which is assessed by the agreement ratio among human judgments. Our best model achieves 0.707 in F-measure and 0.75 in recall rate, and can compress a discussion into 45% of its original duration. The proposed models reveal that nonverbal information is indispensable for selecting meeting extracts of a group discussion. One of the future directions is to implement the models as an automatic meeting summarization system.

This paper focuses on the computational analysis of the individual communication skills of participants in a group. The computational analysis was conducted using three novel aspects to tackle the problem. First, we extracted features from dialogue (dialog) act labels capturing how each participant communicates with the others. Second, the communication skills of each participant were assessed by 21 external raters with experience in human resource management to obtain reliable skill scores for each of the participants. Third, we used the MATRICS corpus, which includes three types of group discussion datasets to analyze the influence of situational variability regarding to the discussion types. We developed a regression model to infer the score for communication skill using multimodal features including linguistic and nonverbal features: prosodic, speaking turn, and head activity. The experimental results show that the multimodal fusing model with feature selection achieved the best accuracy, 0.74 in R2 of the communication skill. A feature analysis of the models revealed the task-dependent and task-independent features to contribute to the prediction performance.

Gaze is an important nonverbal feedback signal in multiparty face-to-face conversations. It is well known that gaze behaviors differ depending on participation role: speaker, addressee, or side participant. In this study, we focus on dominance as another factor that affects gaze. First, we conducted an empirical study and analyzed its results that showed how gaze behaviors are affected by both dominance and participation roles. Then, using speech and gaze information that was statistically significant for distinguishing the more dominant and less dominant person in an empirical study, we established a regression-based model for estimating conversational dominance. On the basis of the model, we implemented a dominance estimation mechanism that processes online speech and head direction data. Then we applied our findings to human-robot interaction. To design robot gaze behaviors, we analyzed gaze transitions with respect to participation roles and dominance and implemented gaze-transition models as robot gaze behavior generation rules. Finally, we evaluated a humanoid robot that has dominance estimation functionality and determines its gaze based on the gaze models, and we found that dominant participants had a better impression of less dominant robot gaze behaviors. This suggests that a robot using our gaze models was preferred to a robot that was simply looking at the speaker. We have demonstrated the importance of considering dominance in human-robot multiparty interaction.

With the goal of assessing participant attitudes and group activities in collaborative learning, this study presents models of participation styles based on co-occurrence patterns of nonverbal behaviors between conversational participants. First, we collected conversations among groups of three people in a collaborative learning situation, wherein each participant had a digital pen and wore a glasses-type eye tracker. We then divided the collected multimodal data into 0.1-second intervals. The discretized data were applied to an unsupervised method to find co-occurrence behavioral patterns. As a result, we discovered 122 multimodal behavioral motifs from more than 3,000 possible combinations of behaviors by three participants. Using the multimodal behavioral motifs as predictor variables, we created regression models for assessing participation styles. The multiple correlation coefficients ranged from 0.74 to 0.84, indicating a good fit between the models and the data. A correlation analysis also enabled us to identify a smaller set of behavioral motifs (fewer than 30) that are statistically significant as predictors of participation styles. These results show that automatically discovered combinations of multiple kinds of nonverbal information with high co-occurrence frequencies observed between multiple participants as well as for a single participant are useful in characterizing the participant's attitudes towards the conversation.

A novel stuffed-toy robot that expresses its intentions through the sense of force and touch is proposed. The robot is so soft that one might wish to embrace it. Compliance of the robot can also be determined by holding its hands, similar to newborn babies or small animals.