• Structural Engineering and Mechanics
• /
• v.76 no.1
• /
• pp.101-114
• /
• 2020

In this paper, an improved experience-based learning algorithm (EBL), termed as IEBL, is proposed to solve the nonlinear hysteretic parameter identification problem with Bouc-Wen model. A quasi-opposition-based learning mechanism and new updating equations are introduced to improve both the exploration and exploitation abilities of the algorithm. Numerical studies on a single-degree-of-freedom system without/with viscous damping are conducted to investigate the efficiency and robustness of the proposed algorithm. A laboratory test of seven lead-filled steel tube dampers is presented and their hysteretic parameters are also successfully identified with normalized mean square error values less than 2.97%. Both numerical and laboratory results confirm that, in comparison with EBL, CMFOA, SSA, and Jaya, the IEBL is superior in nonlinear hysteretic parameter identification in terms of convergence and accuracy even under measurement noise.

• Steel and Composite Structures
• /
• v.42 no.5
• /
• pp.685-697
• /
• 2022

In this paper, a vibration-based method using the change ratios of modal data and the experience-based learning algorithm is presented for quantifying the position, size, and interface layer of delamination in laminated composites. Three types of objective functions are examined and compared, including the ones using frequency changes only, mode shape changes only, and their combination. A fine three-dimensional FE model with constraint equations is utilized to extract modal data. A series of numerical experiments is carried out on an eight-layer quasi-isotropic symmetric (0/-45/45/90)s composited beam for investigating the influence of the objective function, the number of modal data, the noise level, and the optimization algorithms. Numerical results confirm that the frequency-and-mode-shape-changes-based technique yields excellent results in all the three delamination variables of the composites and the addition of mode shape information greatly improves the accuracy of interface layer prediction. Moreover, the EBL outperforms the other three state-of-the-art optimization algorithms for vibration-based delamination detection of composites. A laboratory test on six CFRP beams validates the frequency-and-mode-shape-changes-based technique and confirms again its superiority for delamination detection of composites.

• Journal of Platform Technology
• /
• v.11 no.6
• /
• pp.21-27
• /
• 2023

In this paper, we introduce a reinforcement learning-based algorithm for personalized tourist path recommendations. The algorithm employs a reinforcement learning agent to explore tourist regions and identify optimal paths that are expected to enhance tourism experiences. The concept of tourism experience is defined through points of interest (POI) located along tourist paths within the tourist area. These metrics are quantified through aggregated evaluation scores derived from reviews submitted by past visitors. In the experimental setup, the foundational learning model used to find tour paths is the Deep Q-Network (DQN). Despite the limited availability of historical tourist behavior data, the agent adeptly learns travel paths by incorporating preference scores of tourist POIs and spatial information of the travel area.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.5
• /
• pp.297-307
• /
• 2022

The simple procedural segment selection algorithm commonly used in Dynamic Adaptive Streaming over HTTP (DASH) reveals severe weakness to provide high-quality streaming services in the integrated mobile networks of various wired and wireless links. A major issue could be how to properly cope with dynamically changing underlying network conditions. The key to meet it should be to make the segment selection algorithm much more adaptive to fluctuation of network traffics. This paper presents a system architecture that replaces the existing procedural segment selection algorithm with a deep reinforcement learning algorithm based on the Asynchronous Advantage Actor-Critic (A3C). The distributed A3C-based deep learning server is designed and implemented to allow multiple clients in different network conditions to stream videos simultaneously, collect learning data quickly, and learn asynchronously, resulting in greatly improved learning speed as the number of video clients increases. The performance analysis shows that the proposed algorithm outperforms both the conventional DASH algorithm and the Deep Q-Network algorithm in terms of the user's quality of experience and the speed of deep learning.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.5
• /
• pp.2277-2298
• /
• 2019

Adaptive learning in e-learning has garnered researchers' interest. In it, learning resources could be recommended automatically to achieve a personalized learning experience. There are various ways to realize it. One of the realistic ways is adaptive learning path recommendation, in which learning resources are provided according to learners' requirements. This paper summarizes existing works and proposes an innovative approach. Firstly, a learner-centred concept map is created using graph theory based on the features of the learners and concepts. Then, the approach generates a linear concept sequence from the concept map using the proposed traversal algorithm. Finally, Learning Objects (LOs), which are the smallest concrete units that make up a learning path, are organized based on the concept sequences. In order to realize this step, we model it as a multi-objective combinatorial optimization problem, and an improved immune algorithm (IIA) is proposed to solve it. In the experimental stage, a series of simulated experiments are conducted on nine datasets with different levels of complexity. The results show that the proposed algorithm increases the computational efficiency and effectiveness. Moreover, an empirical study is carried out to validate the proposed approach from a pedagogical view. Compared with a self-selection based approach and the other evolutionary algorithm based approaches, the proposed approach produces better outcomes in terms of learners' homework, final exam grades and satisfaction.

• Journal of the Korea Society of Computer and Information
• /
• v.20 no.5
• /
• pp.151-159
• /
• 2015

The algorithm in computer science includes skills to design a problem solving process for solving problems efficiently and effectively. Therefore all learners who learn computer science have to learn algorithm. Education for algorithm is effective when learners acquire skills to design algorithm as well as ability to use appropriate design skills solving problems. Especially since it is heightened people awareness to cultivating informatics gifted students who have potential of significant impact on society, many studies on how to teach them have been in progress. Therefore in this study we adopted puzzles to help informatics gifted students learn skills to design algorithm and how to use them to solve problems. The results of pre and post test compared to traditional algorithm learning, we identified that puzzled based algorithm learning gave a positive impact to students. Students had various problem solving experience applying algorithm design skills in puzzle based learning. As a result, students of learning and learning transfer has been improved.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10a
• /
• pp.1072-1077
• /
• 1992

In ths paper a new active assembly algorithm for chamferless precision parts mating, is considered. The successful assembly task requires an extremely high position accuracy and a good knowledge of mating parts. However, conventional assembly mehtod alone makes it difficult to achieve satisfactory assembly performance because of the complexity and the uncertainties of the process and its environments such as imperfect knowledge of the parts being assembled as well as the limitation of the devices performing the assebled as well as the limitation of the devices performing the assembly. To cope with these problems, a self-learning rule-based assembly algorithm is proposed by intergaring fuzzy set theory and neural network. In this algortihm, fuzzy set theory copes with the complexity and the uncertainties of the assembly process, while neural network enhances the assembly schemen so as to learn fuzzy rules form experience and adapt to changes in environment of uncertainty and imprecision. The performance of the proposed assembly algorithm is evaluated through a series of experiments. The results show that the self-learning fuzzy assembly scheme can be effecitively applied to chamferless precision parts mating.

• Clinical Endoscopy
• /
• v.57 no.3
• /
• pp.283-292
• /
• 2024

We developed three e-learning systems for endoscopists to acquire the necessary skills to improve the diagnosis of early gastric cancer (EGC) and demonstrated their usefulness using randomized controlled trials. The subjects of the three e-learning systems were "detection", "characterization", and "preoperative assessment". The contents of each e-learning system included "technique", "knowledge", and "obtaining experience". All e-learning systems proved useful for endoscopists to learn how to diagnose EGC. Lecture videos describing "the technique" and "the knowledge" can be beneficial. In addition, repeating 100 self-study cases allows learners to gain "experience" and improve their diagnostic skills further. Web-based e-learning systems have more advantages than other teaching methods because the number of participants is unlimited. Histopathological diagnosis is the gold standard for the diagnosis of gastric cancer. Therefore, we developed a comprehensive diagnostic algorithm to standardize the histopathological diagnosis of gastric cancer. Once we have successfully shown that this algorithm is helpful for the accurate histopathological diagnosis of cancer, we will complete a series of e-learning systems designed to assess EGC accurately.

• Journal of IKEEE
• /
• v.25 no.2
• /
• pp.337-343
• /
• 2021

Deep reinforcement learning is an artificial intelligence algorithm that enables learners to select optimal behavior based on raw and, high-dimensional input data. A lot of research using this is being conducted to create an optimal movement path of a mobile robot in an environment in which obstacles exist. In this paper, we selected the Dueling Double DQN (D3QN) algorithm that uses the prioritized experience replay to create the moving path of mobile robot from the image of the complex surrounding environment. The virtual environment is implemented using Webots, a robot simulator, and through simulation, it is confirmed that the mobile robot grasped the position of the obstacle in real time and avoided it to reach the destination.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.