Personalized Federated Learning Algorithm with Adaptive Clustering for Non-IID IoT Data Incorporating Multi-Task Learning and Neural Network Model Characteristics.

The proliferation of IoT devices has led to an unprecedented integration of machine learning techniques, raising concerns about data privacy. To address these concerns, federated learning has been introduced. However, practical implementations face challenges, including communication costs, data and device heterogeneity, and privacy security. This paper proposes an innovative approach within the context of federated learning, introducing a personalized joint learning algorithm for Non-IID IoT data. This algorithm incorporates multi-task learning principles and leverages neural network model characteristics. To overcome data heterogeneity, we present a novel clustering algorithm designed specifically for federated learning. Unlike conventional methods that require a predetermined number of clusters, our approach utilizes automatic clustering, eliminating the need for fixed cluster specifications. Extensive experimentation demonstrates the exceptional performance of the proposed algorithm, particularly in scenarios with specific client distributions. By significantly improving the accuracy of trained models, our approach not only addresses data heterogeneity but also strengthens privacy preservation in federated learning. In conclusion, we offer a robust solution to the practical challenges of federated learning in IoT environments. By combining personalized joint learning, automatic clustering, and neural network model characteristics, we facilitate more effective and privacy-conscious machine learning in Non-IID IoT data settings.

A study of students' learning perceptions and behaviors in remote STEM programming education.

In recent years, STEM education has developed students' fundamental subject knowledge, and has allowed students to integrate STEM cross-domain knowledge. Universities emphasize innovative thinking, practice and application, problem-solving, and teamwork to nurture students while learning STEM cross-domain knowledge development in remote STEM program education. When students take online STEM programs, they may encounter unanswered questions and may give up on trying to solve them. Therefore, this study proposed a problem-based learning approach with an online programming system integrated into an online STEM programming course. To help students solve the past programming assignments, the assignments were simplified, decomposed, and reorganized. The teacher guided the students to understand the STEM programming problems and taught them to use appropriate problem-solving skills to motivate them to complete the STEM programming assignments. The experiment was conducted with students in an online STEM programming course at a university in northern Taiwan. In the experimental activities, we used a problem-based learning approach for the online STEM programming activity. The problem-based learning method can be divided into four steps, namely stating the problem, understanding the problem, developing a solution plan, and executing the plan, reflecting, and debugging. This study used a problem-based learning approach and an online programming system integrated into a STEM programming curriculum to explore the differences in students' perceptions of STEM learning, learning outcomes, and learning behaviors. The experimental results found a significant difference between students' prior knowledge and learning outcomes. Students showed significant gains in learning the STEM programming content using the problem-based learning approach and the online programming system. In the analysis of their STEM learning perceptions, we found that there were significant differences in students' responses for each dimension. This shows that using the problem-based learning approach with the online programming system helped students learn the course content. The analysis of students' behaviors in answering the STEM programming assignments indicated that some students had the habit of taking notes. This helped them to easily associate and integrate STEM cross-domain knowledge with what they had learned in the online course, and enhanced their ability to implement STEM programs. In addition, students could take the initiative and focus on repeatedly watching the teacher solve the material in the online course. Students could try different solving plans to pass the code validation of the STEM programming assignments. This revealed that students wanted to complete the STEM programming assignments to achieve good learning performance.

The effects of applying an augmented reality English teaching system on students' STEAM learning perceptions and technology acceptance.

With the inherently interdisciplinary and hands-on nature of the STEAM system, this study changes the way the English language is taught, no longer confining it to monotonous tests and dull textbooks. Teaching via Augmented Reality (AR) technology tends to pique students' interest, making them more willing to participate in learning. This study used AR technology in conjunction with gaming, where the participants were split into pairs, which promoted friendly competition among themselves, and, in turn, drove them to learn more. Moreover, the ability to track their own performance was added to the system, as well as an in-depth explanation for every question. The teacher used our teaching aids system in the lessons to make the teaching process more interesting and enjoyable. There were two English course experiments, each lasting for 4 weeks. The goal of the first experiment was to gain a general understanding of how students felt about the lessons and how well they could operate the system. After analyzing the results, some improvements for problems that arose were made in the second experiment. The goal of the second experiment was to find out how students felt about the lesson content and teaching aids and to compare the results of the two experiments. To help us find potential oversights and check how participants felt about this approach, surveys regarding STEAM learning and overall ease of use were administered. The results showed that AR English learning was widely considered helpful by the students, and also, our system was easy to operate, and competing with peers really motivated them to do better.

The Influence of Affective Feedback Adaptive Learning System on Learning Engagement and Self-Directed Learning.

The outbreak of the two-year corona virus has made a great difference on existing methods of learning and instruction. Online education has become a crucial role to maintain non-stop learning after the post-epidemic period. The advanced technologies and growing popularity of network equipment have made it easy to deploy remote connections. However, teachers still face challenges when they actually implement distance courses. During the learning process, the quality of learning can be improved if the researchers consider multiple factors, including emotions, attitudes, engagement, cognition, neuroscientific and cultural psychology. After analyzing these factors, instructors can have better understanding of students' mental building and cognitive understanding in their process of learning, and be familiar with the way of interaction with students and appropriately adjust their teaching. Therefore, the current study established a learning system that aimed to understand learners' emotional signals during learning by applying the adaptive-feedback emotional computing technology. The purpose of the system was to allow learners to (1) self-examine their learning condition, (2) enhance their self-directed learning, (3) help learners who are in negative learning emotions or settings to lower anxieties, and (4) promote their learning attitudes and engagement. Result showed that the system with the adaptive-feedback emotional computing technology has significantly improved the learning effectiveness, lowered learning anxieties and increased students' self-directed learning.

Study of Virtual Reality Immersive Technology Enhanced Mathematics Geometry Learning.

Mathematics is an important foundation for the development of science education. In the past, when instructors taught mathematical concepts of geometry shapes, they usually used traditional textbooks and aids to conduct teaching activities, which resulted in students not being able to understand the principles completely. Nowadays, it has become a trend to integrate emerging technologies into mathematics courses and to use digital instructional aids. Emerging technologies can effectively enhance students' sensory experience while strengthening their impressions and understandings of subject concepts. In this paper, we apply virtual reality immersive technologies to develop a "virtual reality immersive learning mathematics geometry system," which is used to teach mathematical geometry concepts. Teachers use the system to develop three basic mathematical geometry learning materials: "Triangular pyramid volume = 1/3 prism volume," "Cone volume calculation," and "Triangle center of gravity derivation." In the experimental activity, the teacher uses virtual reality teaching aids to guide students to learn mathematical geometry concepts in a fun way so that they can achieve the effectiveness of immersive learning. This study explores the impact of using the virtual reality immersive learning mathematics geometry system on students' technology acceptance, learning motivations, and learning performance. The experimental result showed that using the virtual reality immersive learning mathematics geometry system can improve the learning motivation and learning performance of students. The findings indicated that the experimental group had better learning outcomes after completing the learning tasks of three geometric units. The experimental group used the virtual reality immersive learning mathematics geometry system which can lead to better learning outcomes. According to the ARCS questionnaire, students in the experimental group were confident to understand new subjects. At the same time, the mode of completing the game can effectively give students a sense of accomplishment. The use of emerging technologies in the classroom can be an attractive learning mode for students.

Applying Educational Data Mining to Explore Viewing Behaviors and Performance With Flipped Classrooms on the Social Media Platform Facebook.

In recent years, learning materials have gradually been applied to flipped classrooms. Teachers share learning materials, and students can preview the learning materials before class. During class, the teacher can discuss students' questions from their notes from previewing the learning materials. The social media platform Facebook provides access to learning materials and diversified interactions, such as sharing knowledge, annotating learning materials, and establishing common objectives. Previous studies have explored the effect of flipped classrooms on students' learning engagement, attitudes, and performance. In this paper, we apply educational data mining to explore the relationship between students' viewing behaviors in accessing learning materials and their performance in flipped classrooms. The participants are classified into an experimental group and a control group to engage in flipped classroom activities. The experimental group uses the social media platform Facebook for flipped learning, and the control group uses a learning management system for flipped learning. The results show that there is a significant difference in the learning performance between the two groups, with the average score of the experimental group being higher than that of the control group. Furthermore, we find that the viewing behaviors and performance of the students within the experimental group differ significantly.

To Cultivate Creativity and a Maker Mindset Through an Internet-of-Things Programming Course.

With the swift development of technology in recent years, entrepreneurs are facing rapid changes in industry. To cope with such changes at home and abroad, The Ministry of Education is actively promoting innovative education with the aim of cultivating students' entrepreneurship. On this basis, this study proposes an innovative curriculum design based on an Internet-of-Things (IoT) programming course. The reason is that it develops computational thinking skills while students are learning programming and also cultivates logical thinking skills and problem-solving skills, which are critical to entrepreneurship. We also design a number of learning activities that enable students to express their opinions and ideas while gaining more knowledge through peer interaction and discussion. Overall, this study explores the impact of "maker education" on students' attitudes toward computer thinking. The results indicate that maker education has a positive impact on their ability to learn computer skills. In terms of learning motivation, students are not motivated by maker education and reduce their confidence on the curriculum. The reason may be that the curriculum requires the acquisition of software and hardware skills, which will increase the student's learning burden, so they more likely to encounter learning disabilities.

An Intelligent Body Posture Analysis Model Using Multi-Sensors for Long-Term Physical Rehabilitation.

Sensors can be installed on various body parts to provide information for computer diagnosis to identify the current body state. However, as human posture is subject to gravity, the direction of the force on each limb differs. For example, the directions of gravitational force on legs and trunk differ. In addition, each person's height and structure of limbs differs, hence, the acceleration and rotation resulted from such differences on force and length of the limbs of a person in motion would be different, and be presented by cases of different postures. Thus, how to present body postures through skeleton system equations, and achieve an long-term physical rehabilitation, according to the different limb characteristics of each person, is a challenging research issue. This paper proposes a novel scheme named as "Intelligent Body Posture Analysis Model", which uses multiple acceleration sensors and gyroscopes to detect body motion patterns. The effectiveness of the proposed scheme is proved by conducting a large number of practical experiments and tests.

An Adaptive Sensor Data Segments Selection Method for Wearable Health Care Services.

As cloud computing and wearable devices technologies mature, relevant services have grown more and more popular in recent years. The healthcare field is one of the popular services for this technology that adopts wearable devices to sense signals of negative physiological events, and to notify users. The development and implementation of long-term healthcare monitoring that can prevent or quickly respond to the occurrence of disease and accidents present an interesting challenge for computing power and energy limits. This study proposed an adaptive sensor data segments selection method for wearable health care services, and considered the sensing frequency of the various signals from human body, as well as the data transmission among the devices. The healthcare service regulates the sensing frequency of devices by considering the overall cloud computing environment and the sensing variations of wearable health care services. The experimental results show that the proposed service can effectively transmit the sensing data and prolong the overall lifetime of health care services.

A collaborative computing framework of cloud network and WBSN applied to fall detection and 3-D motion reconstruction.

As cloud computing and wireless body sensor network technologies become gradually developed, ubiquitous healthcare services prevent accidents instantly and effectively, as well as provides relevant information to reduce related processing time and cost. This study proposes a co-processing intermediary framework integrated cloud and wireless body sensor networks, which is mainly applied to fall detection and 3-D motion reconstruction. In this study, the main focuses includes distributed computing and resource allocation of processing sensing data over the computing architecture, network conditions and performance evaluation. Through this framework, the transmissions and computing time of sensing data are reduced to enhance overall performance for the services of fall events detection and 3-D motion reconstruction.

An environmental-adaptive fall detection system on mobile device.

When facing damages caused by falls, a well designed smart sensor system to detect falls can be both medically and economically helpful. This research introduces a portable terrain adaptable fall detection system, by placing accelerometers and gyroscopes in parts of the body and transmit data through wireless transmitter modules to mobile devices to get the related information and combining it with the center of gravity clustering algorithm introduced in this research which computes the human body behavior patterns according the relationship between the center of gravity in the body and the feet portion of the body. Compared with the research in the past, this system is not only highly accurate and robust, but also able to adapt to different types of terrains, which solves the problems that other researches have for detection errors when the client is climbing the stairs or walking on a slant.