Conference Papers

In view of the mental health issues among adolescents in Singapore, aromatherapy is proposed to mitigate their stress level when studying. An experiment was conducted both in the classroom and home setting, to test the effectiveness of the give odours, lavender, rosemary, ylang-ylang, lemon and bergamot on reducing stress level, using the no odour scenario as control. Objective data (SpO2, heart rate and stress score) is collected using the HUAWEI Band 6 smartwatch. Subjective data was self-reported by the participants through Google forms, rating their emotional health on a 10-point scale and elaborating in prose. It was found that the anti-anxiety effects of the stimulants (lemon, bergamot, rosemary) were much larger than that of the sedatives (lavender, ylang-ylang). In particular, lemon showed the best objective anxiolytic effect, while bergamot was the best in terms of self-perceived effect. Rosemary relieved stress through raising productivity, but some effects of overworking were observed. On the other hand, ylang-ylang showed inconsistent effects, while lavender was not suitable to relieve stress when studying.

This research investigates distinct human-generative AI collaboration types and students’ interaction experiences when collaborating with generative AI (i.e., ChatGPT) for problem-solving tasks and how these factors relate to students’ sense of agency and perceived collaborative problem solving. By analyzing the surveys and reflections of 79 undergraduate students, we identified three human-generative AI collaboration types: even contribution, human leads, and AI leads. Notably, our study shows that 77.21% of students perceived they led or had even contributed to collaborative problem-solving when collaborating with ChatGPT. On the other hand, 15.19% of the human participants indicated that the collaborations were led by ChatGPT, indicating a potential tendency for students to rely on ChatGPT. Furthermore, 67.09% of students perceived their interaction experiences with ChatGPT to be positive or mixed. We also found a positive correlation between positive interaction experience and a sense of positive agency. The results of this study contribute to our understanding of the collaboration between students and generative AI and highlight the need to study further why some students let ChatGPT lead collaborative problem-solving and how to enhance their interaction experience through curriculum and technology design.

The aim of this study is to examine the online learning experiences of university students with Special Educational Needs (SEN), and how their experiences might differ from their typically developing peers. Fifty typically developing students (mean age = 22; 29 females) and 31 students with SEN (mean age = 22; 15 females) from a local university in Singapore participated in an online survey. Both groups reported significant increase in the proportion of online learning after the outbreak of COVID-19 pandemic. Both groups reported being moderately positive about their online learning experiences, with no significant difference between the groups (either before or after the outbreak). For both groups, Learning Activity Management System (LAMS), pre-recorded lectures, online finals/quizzes, live lectures, online assignments, and online tutorials were the common online learning formats. Laptop/desktop was the primary device used, and Zoom was the most preferred online learning software. The SEN group reported higher usage of technical accommodations. Accessibility was the top advantage of online learning reported by typically developing students while for students with SEN, it was flexibility. Lower social interaction was the top challenge encountered for both groups. These findings would be useful in making online learning more inclusive for everyone in university.

Blended synchronous learning (BSL) is an instructional approach that enables online students to participate in classroom activities from geographically separated sites using video conferencing technologies. Despite its educational benefits, maintaining and increasing the engagement of online students is challenging. In this study, some strategies were adopted in two classes (N=22 & 23) to investigate how online students could be effectively engaged and their perceptions of the strategies applied. Surveys and focus group discussions were administered. Results showed that leading group discussions was helpful for online students to be engaged. However, it had challenges for online students as they did not know who was talking and not every member could be observed in the video. Having a teaching assistant (TA) was highly rated. It enabled the instructor to pay close attention to the questions posted to the chat box promptly and helped online students know what was happening in class when the connection was unstable. Giving peer feedback was another useful strategy. However, it only worked when everyone was familiar with the assignment topics of others. Using an interactive tool like Pear Deck did not noticeably increase student engagement. It seemed the design of learning content and activities was more important than the tool itself. In addition, the students commonly indicated that they were highly engaged, and they did not think that their engagement level was lower when they were online. This finding was inconsistent with existing literature, which requires further investigation in the future. Implications for practitioners and researchers are discussed.

Fitness exercises, including push-ups, are very beneficial to personal health. Many Artificial Intelligence (AI)-based fitness trainers are developed based on human pose estimation models or assisted by Internet of Things (IoT) devices. However, many of them require access to a graphing processing unit (GPU) for model training or IoT sensors to deploy, less accessible for individuals. In our work, we designed and prototyped real-time mobile push-up detectors using three distinctive approaches: (1) Push-up pose classification, (2) Angle-heuristic estimation and (3) Optical flow detection. We trained our deep-learning model with over 2000 images to achieve a high accuracy for real time deployment. Models are tested on our video dataset applied data augmentation techniques to simulate real-world environmental conditions to evaluate model performance based on accuracy metrics (precision, recall, F1 score) and processing frame rate (FPS). From the results, we concluded that the angle-heuristic estimation method has the best overall performance and we analysed the reasons for the relatively poorer performance of the push-up pose classification and optical flow detection methods. All methods developed are capable of working on mobile devices without the need of GPU or IoT sensors.