Children with intellectual disabilities (ID), especially those who fail to communicate with normal speech, suffer from various limitations in their social functioning. They often rely on alternate methods to supplement or replace verbal communication. There is a rapidly growing number of augmentative and alternative communication (AAC) mobile applications available over recent years. However, the operation of these applications remains a challenge for users with ID since they may not have the sufficient cognitive proficiency to cope with the application interface. In this paper, we present the design, implementation, and evaluation of a Bluetooth low energy (BLE) based context-aware AAC system. Our system facilitates daily communication for nonverbal school children with moderate ID in order to address their learning and communication needs. We have piloted and evaluated our system at a special education school in Hong Kong. Our technical contribution is twofold. First, we introduce BLE technology to the AAC field and show its effectiveness on users with ID. Second, by converting conventionally standalone AAC software into a service system and empowering it with Internet of Things functionality, we open up many new opportunities brought by the system-of-systems (SoS) paradigm for the intellectually disabled population.

Affective computing focuses on recognizing emotions using a combination of psychology, computer science, and biomedical engineering. With virtual reality (VR) becoming more widely accessible, affective computing has become increasingly important for supporting social interactions on online virtual platforms. However, accurately estimating a person’s emotional state in VR is challenging because it differs from real-world conditions, such as the unavailability of facial expressions. This research proposes a self-training method that uses unlabeled data and a reinforcement learning approach to select and label data more accurately. Experiments on a dataset of dialogues of VR players show that the proposed method achieved an accuracy of over 80% on dominance and arousal labels and outperformed previous techniques in the few-shot classification of emotions based on physiological signals.

Nowadays people's life become more rich and convenient than before, but the obesity also become a problem with people. One cause of obesity is lack of exercise. Although the reason is known by people and the method of keeping health is self maintenance, but it is hard for most people, especially the elderly people. Because of this, we design a daily exercise support application, which is based on the user model, to provide health maintain support. In this research, we focus on applying user model to help people. Because a person's health situation relates to his everyday activities, the users' data of daily life can be the data to build the user model. Thus we choose to use sensors to recode the amount of exercise and users' motion data to build the user model. We obtained the user's motion data and the amount of exercise from sensors firstly, and use fuzzy inference and fuzzy membership function to recognize the motion data. And then, the user model can be constructed by mining motion data and the amount of exercise. Finally, the daily exercise support application can offer service to the user by applying the user model. Furthermore, we will show two kind of exercise support service of the application.

Title: Unsupervised Anomaly Detection for IoT-Driven Multivariate Time Series on Moringa Leaf Extraction | Keywords: moringa leaf extraction, IoT, unsupervised anomaly detection, multivariate time series | Author: Kurnianingsih, Retno Widyowati, Achmad Fahrul Aji, Eri Sato-Shimokawara, Takenori Obo, and Naoyuki Kubota

Title: The Influence of Robot’s Expressions on Self-Efficacy in Erroneous Situations | Keywords: self-efficacy, Laban movement, robot expression, human-robot interaction | Author: Youdi Li, Haruka Sekino, Eri Sato-Shimokawara, and Toru Yamaguchi