There is growing interest in communication artifacts that allow users to connect with others while supporting the emotional well-being that often arises from social interactions. Social presence is central in digital interactions, as it fosters a sense of “being together,” which is essential for affective communication, collaboration, and relationship-building in computer-mediated environments. This study explores the impact of behavioral synchronization through HugBits, a cushion -shaped device designed for remote hug-based interaction, on both perceived social presence and users' prosociality. We conducted two experiments: (1) testing whether hug synchronization increased perceived social presence, and (2) assessing whether synchronization influenced cooperative decision-making in a Prisoner's Dilemma game. In both experiments, the occurrence of behavioral synchronization (i.e., hugging) was controlled by having participants interact with a virtual agent that was designed to probabilistically promote or avoid synchronized hugs. Results show that more synchronized interactions significantly enhanced perceived social presence and promoted cooperative behavior. These findings highlight behavioral synchronization as a promising design strategy for affective communication technologies and provide concrete implications for integrating emotional and behavioral outcomes in computer-mediated communication research.

Preference learning is often used to adjust the output of LLM after instruction tuning for more desirable outputs. However, since creating datasets for preference learning is challenging, we are considering automatic creation based on user history. As a new method for label estimation in this context, we propose a similar sample search method for preference learning datasets, which compares data with similar labeled samples. This method is based on the relational structure of preference learning datasets, where data are organized as pairs of Chosen and Rejected.

Concept drift, defined as a change in the conditional distribution of the target given the features, can seriously degrade the predictive performance of machine-learning models. Existing analysis methods often lack intuitive visual summaries, making it difficult to understand and address drift. We introduce CDST-Viz, a framework for the segmentation and visualization of concept drift. CDST-Viz first trains a Concept Drift Segmentation Tree (CDST) that partitions the feature space into regions with similar drift patterns across two time periods. The trained tree is rendered into an interactive treemap and accompanied by time-series plots of the target in each region to reveal where and how the distribution shifts. Case studies on six public regression datasets show that CDST-Viz clearly reveals fine-grained drift patterns and supports informed responses to concept drift.

Ensemble learning that combines multiple weak learners for enhanced performance, is widely used but suffers from low interpretability/explainability. This leads challenges not only in operational aspects like model maintenance and quality assurance but also in addressing societal needs such as fairness and privacy. To tackle this, we propose a new visualization method focusing on the relationship among weak learners in ensemble models to improve understanding of the model structure and its learning processes. In this paper, we defined the relation between weak learners based on a “common sample” in gradient-boosting decision trees, and a visualization method as a three-dimensional graph structure was proposed. Ensemble models trained with synthetic data sets that include typical distribution shifts and real-world open data sets were visualized. As a result, we demonstrated that this approach enables a more accessible understanding of the behavior and structure of ensemble models comprising multiple weak learners, facilitating the identification of overfitting and underfitting through visualization of changes during the training and validation processes.

Understanding temporal changes in data is crucial for successful MLOps, however, this understanding is generally challenging. This paper presents an algorithm that visualizes multiple data sets and a prediction model in a single plot. The proposed algorithm efficiently captures the changes in characteristics of the data sets based on accuracies of models trained on each data set. Our visualization enables data scientists to easily identify trends in the direction of change, periodicity, and anomalies in the data sets, as well as the predictive performance of the prediction model, all at a glance. Case studies using six real-world open data demonstrates that our visualization effectively provides valuable insights into the changes of the data sets, facilitating more informed decision-making in MLOps.

Live Performances have been conducted online in recent years due to the influence of the COVID-19 pandemic. This generated a novel problem of losing eye contact with the performers, which deteriorates the viewing experience in the online setting. This study proposes a feedback system based on gaze modeling to augment the perceptual experience of being faced by the performer during an online live performance viewing. This system aims to improve the experience of viewing online live performances by recreating a perceptual experience of interaction between the performer and the audience. The proposed system consists of a gaze recognition unit that judges whether the performer is looking at the camera based on gaze modeling in the video, and a stimulus presentation unit that presents the judgement result to the audience through vibration stimulus. This paper describes the gaze recognition model, experiments to determine the parameters to be used in the model, and the evaluation of live performance viewing experiments using the proposed system.

In this study, we developed a smart mirror device for continuous mental health checkup based on individual response characteristics through simple and short dialogue interaction on a daily basis. Early detection and intervention have a predominant impact on remission in many cases of mental illness, and daily monitoring is essential for early detection. However, existing methods require time-consuming and burdensome measurements and rely on information obtained from patients' experiences, communication, and physical findings for diagnosis. We propose a mental health checkup system that uses supervised learning to model physician's evaluation based on the response characteristics of individuals during short dialogue interaction with the smart mirror. A validation experiment with 10 participants was conducted for two weeks, and the results showed that mental health checkups by physician was successfully reproduced with the proposed algorithm. In conclusion, this study demonstrates the potential of Smart Mirror for mental health checkup in daily life. It provides a potentially more objective and non-invasive method for early detection and monitoring, which could contribute to gradual improvements in mental healthcare.

The amount of force and the timing of force application are necessary training skills in basketball free-throw practice. Coaches cannot readily assess the shooter's motion and provide instructions because the throwing motion is performed in a short time, and because the applied force cannot be directly observed by a second person. In this work we propose a wearable device that measures the wearer's force control during throwing based on electrical impedance and acceleration measurement of the forearm, and provides feedback such that the wearer can adjust their motion appropriately. Temporal features of the electrical impedance and acceleration measured during throwing are determined experimentally to differentiate between experts and beginners. Using these features, we propose a training system for force control using the proposed device and verify its effectiveness in free-throw training.

International migration forces people into an unfamiliar reality in which their customs and values lose relevance. Moreover, former relationships are left behind, which makes immigrants more likely to experience loneliness. This study focuses particularly on Venezuelan immigrants by incorporating cultural aspects into a solution aimed at reducing loneliness and increasing social connectedness. Among Venezuelans, coffee is a staple of their daily routine and their favorite social beverage. We propose KEPEIN, a coffee maker-shaped interface to transfer a sense of presence and share coffee over distance. Through an experimental study, we evaluated the user’s perception and reaction when communicating through the interface. The results show potential added value to communication by including KEPEIN in a traditional remote interaction scenario. We discuss the benefits and limitations of this type of tangible communication interface and the importance of incorporating culture into the design of solutions for immigrants.

As the operation of robots is becoming easier, adopting a robot as a companion for families is increasing. When considering that family playtime has a positive influence on all family members, it is important to provide various activities for families to play together. Robots might mediate fun activities to increase family playtime. Therefore, this research was designed to compare play activities between dyadic child-robot interactions and family-robot interactions. We explored the possible activities for family playtime with a NAO robot. The results from three family groups showed that verbal activities, such as giving an instruction to a robot and talking to a robot, increased when family members are playing together with the robot. Also, increased physical activities were involving all family members, such as giving and receiving balls. Therefore, it will be necessary for a robot to recognize the voices of multiple family members and respond to each member appropriately. More importantly, a robot's time allocation for each family member and encouraging the participation of all members will be necessary to maintain or Increase the family playtime.

We propose a method for learning transferable skills among various state-action spaces for reinforcement learning using embedding spaces. Most existing approaches for robot skill transfer assume similar hardware settings between robots. However, in practical applications, identical hardware settings among different robots is not necessarily guaranteed. This study is an attempt to abstract and represent the characteristics of state-action space caused by differences in hardware, such as sensor settings and actuators, using an embedding space. Using this method, the skills become transferable between different hardware settings without requiring heuristic tuning by experts. In this study, we implement the proposed method on a mobile robot in a two-dimensional plane and demonstrate its effectiveness by performing a navigation task in a simulation environment.

In computer-mediated communication, the amount of non-verbal cues or social signals that machines can support is still limited. By integrating haptic information into computational systems, it might be possible to give a new dimension to the way people convey social signals in mediated communication. This research aims to distinguish different haptic gestures using a physical interface with a cushion-like form designed as a mediator for remote communication scenarios. The proposed interface can sense the user through the cushion's deformation data combined with motion data. The contribution of this paper is the following: 1) Regardless of each participant's particular interpretation of the gesture, the proposed solution can detect eight haptic gestures with more than 80% of accuracy across participants, and 2) The classification of gestures was done without the need of calibration, and independent of the orientation of the cushion. These results represent one step toward the development of affect communication systems that can support haptic gesture classification.

This study aims for the new driving skill improvement support system based on driving skill evaluation. We try to define driving skill in terms of "spatial perception" and "prediction". This research can be roughly divided into two steps. The first step is to establish an objective skill evaluation model. The second step is to verify effects of the support system for driving skill improvement. In this paper, we describe that we evaluated driving skill based on parking performance, operational behavior like steering and cognitive behavior like head motion and gazing points. We conducted driving behavior measurement experiment by using a real vehicle to get operational and cognitive data during driving. In consequence, we confirmed that driving skill could be defined by spatial perception and operational behavior of drivers in approaching to an obstacle.

This paper investigates the effects of real time visual biofeedback for improving sports performance using a large scale immersive mixed reality system in which users are able to play a simulated game of curling. The users slide custom curling stones across the floor onto a projected target whose size is dictated by the user's stress-related physiological measure; heart rate (HR). The higher HR the player has, the smaller the target will be, and vice-versa. In the experiment participants were asked to compete in three different conditions: baseline, with and without the proposed biofeedback. The results show that when providing a visual representation of the player's HR or "choking" in competition, it helped the player understand their condition and improve competition performance (P-value of 0.0391).

Affective communication technologies are designed to enhance awareness, social connectedness, and affectivity. Design strategies involve alternative methods to convey affection in computer-mediated scenarios, emphasizing on the importance of mediated physical contact. Therefore, we proposed a huggable interface to mediate social touch by sensing the user's hug gestures, transferring them to a paired device, and delivering them as simple cues. We investigated the effect of the huggable interface as a mediator with a physical embodiment and compared it with a similar communication interface represented by an agent with a virtual embodiment on a touch screen. During the experiments, we set up a scenario in which individuals with a close relationship watched movies and communicated with each other. Results showed the effect of both interfaces in terms of perceived social connectedness and stress levels. The discussion pointed out the potential and limitations of the proposed evaluation method, as well as of each type of interface as affective communication technology.

It is well-known that walking offers many health benefits for everyone, especially for older people who need to maintain mobility and independence coping with declining of functional capacity. In this paper, we present the design of a humanoid walking trainer that has to monitor and encourage walking in the elderly. This design is based on our target users' preferences. We present as well a preliminary walking experiment that was carried out in order to test the accuracy of the gait data obtained from the laser range sensor, which is positioned on the robot, during motion.

Although socially assistive robots are popularly studied as a powerful tool in therapeutic activities for children with Autism Spectrum Disorder (ASD), there are several challenges particularly in automation of the robot's behavior due to difficulties of sensing children's social behaviors by the robotic agent. We developed measurement methods to quantitatively measure children's social behaviors, such as smiling and facing behavior, by means of a wearable device and an image processing. Smiling, facing behavior and their synchrony are usually considered as important social behaviors that imply positive affect and engagement of the participant during the therapeutic intervention. Therefore, the proposed approach would contribute not only to develop the robotic agent which is capable of performing in response to social behaviors of children with ASD, but also to support an evidence-based therapeutic intervention. In this article, the measurement performance of the proposed methods were verified by experiments with ten children with ASD, and the relationship between their smile and facing behavior were analyzed.

Engaging and motivational toys used during therapies for children with Autism Spectrum Disorder (ASD) need to be combined with the appropriated evaluation methods. Technology can provide the therapists with tools to facilitate the analysis of the interventions. In this paper we propose a model using paired devices with three different interaction rules made to facilitate turn-taking behaviors. These rules are evaluated in succession using identical spherical devices with simple visual cues designed to mediate turn-taking interaction between children and therapists. The results explored the potential of paired devices with embedded sensors as a method not only to engage children in the activity, but also to measure social cues that describe children performance. The evaluation of the proposed interaction model was used to provide insights for future implementations of paired devices for training turn-taking.

This study proposes “Gaming Humanoid”, a robot that is able to play video games with human as a partner in the same gaming conditions found typically in a gaming environment, in particular, using body gestures and being able to adapt the skill level. We developed an autonomous video gameplay capability of the humanoid robot NAO. In the autonomous gameplay, we used image processing to recognize the gaming situation and play along with the content. Also we designed the robot's body movement to perform realistic gameplay and to be able to adapt gaming performance. The result with Tennis Game shows that the gaming humanoid was able to play in real gaming environment and also able to control performance of gameplay. The experiment with human showed that the gaming humanoid was not inferior to human level in terms of gameplay. Moreover, the preliminary study with children demonstrated its feasibility of facilitating interactions by implementing emotional/encouraging behaviors. In this paper, we introduce the gaming humanoid as a playmate partner for human, and we describe and evaluate the whole system.

Among treatments for children with ASD, assistive robots are growing popular as they are able to elicit different social behaviors. At the same time, technology is providing methods to automatically collect quantitative data, in order to assist the therapist evaluating the children progress. In this study we introduce a system composed of multiple spherical devices, as well as the design of a turn taking activity using those devices. In previous studies turn taking was found to be an important social skill for development and engaging in activities with others. To evaluate the system performance, a single case experiment with a boy with ASD and the developed device was done. During the activity, the device had two different roles: to engage the child on the turn taking activity and to provide information to the therapist that describe the child's behavior. We could successfully collect quantitative data that represent turn taking as well as we could observe how does the boy manipulate and interact with the device. Based on the results, we are motivated to keep exploring the potential application of devices mediated activities for children with ASD.

This paper presents the design, implementation, and application of a vision-based automatic system that measures facial interaction based on human's cognitive feature. We investigated the feature of people's facial interaction under natural gaze-movement via an experiment, and created criteria of k-degree facial interaction and face-to-face interaction depending only on facial orientation. These criteria were used to develop the automatic system. The system is vision-based. It could be easily embedded into applications. We focused on an application to understand the behavior of children with autism spectrum disorders (ASD), and tested the use of the automatic system in a robot-assisted activity for those children. The results suggested that the system could help to improve the efficiency of behavior analysis during the children's activities with the robot. The facial interaction measured between the children and the robot can be used by therapists to comprehend the children's psychological aspects and state of health.

This paper introduces a doll-type interface for real-time teleoperation of a humanoid robot in Robot-Assisted Activity (RAA) for children with Autism Spectrum Disorders (ASD). We developed a prototype of the interface and verified its usability by conducting RAA sessions with children with ASD.

Recent studies on autism spectrum disorders (ASD) have reported that positive emotions can be a good incentive for children with ASD to perform spontaneous positive social behaviors. Based on this findings, we propose an affective robot-assisted activity (ARAA) for fostering social interaction and communication skills among children with ASD by promoting their positive emotional responses through interaction with a robot. As it has been termed spectrum, every child has different social and affective characteristics that should be taken into account. However, due to difficulties in programming the robot's behavior, conventional RAA systems did not allow the therapist to customize the activity according to the characteristics of each individual. To tackle this problem, we developed a comprehensive framework of ARAA that consists of (i) a robot tele-operation method that allows a therapist to improvise a robot's behavior in real-time and (ii) a quantitative measurement method to describe social interaction within both behavioral and affective aspects.

This study proposes a gaming companion robot to facilitate social interaction among children. Games have been one popular social communication tool for people, and we propose a novel approach to gaming communication using a robot. We have implemented a robot that is capable of playing a video game to facilitate social interaction in a game environment. We used autonomous control to allow the robot to play during the videogame, and the Wizard of OZ (WOZ) framework for interaction between human and robot before and after the game. The results show that the robot was able to play the game in a manner similar to how a human plays the game. The preliminary study with children showed that this new approach was useful to facilitate people's interactions. We describe and evaluate the robotic system in this paper.

The accident probability of beginner drivers is significantly higher than that of experienced drivers. It can be assumed that this is due to lack of driving skills which lead to making wrong decisions according to cognition and operating in correct way. In this paper, we propose a novel assisted driving system intended to help drivers to improve their skills for the reverse parking. The system is able to assist the driver by haptic instruction on the steering wheel in order to induce the driver to make the adequate operation. For the validation, we developed a 1/10 scale car simulator as a simulation environment on which we installed the proposed assistance method and conducted reverse parking experiment by using the simulator. According to the experiment, we validated that the parking accuracy and the trajectory similarity of subjects assisted by proposed system significantly increased compare to subjects unassisted. Consequently, the proposed assisted driving system could accelerate the learning of humans' driving skills.