Author name cluster

Jeng-Shyang Pan

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

1 author row

EAAI Journal 2026 Journal Article

Large language model-driven dynamic communication strategy generation for multi-swarm particle swarm optimization

Tongbang Jiang
Shu-Chuan Chu
Jeng-Shyang Pan
Václav Snášel
Yicheng Wei

Multi-swarm particle swarm optimization (MSO) has emerged as a crucial approach to enhance the search efficiency and global performance of swarm intelligence, showing broad applicability in complex problems. However, existing MSO methods generally rely on predefined communication strategies, which are insufficient to adapt to dynamic changes across different optimization stages and problem characteristics, extremely limiting their flexibility and generalization. To address this issue, this paper proposes a large language model-driven dynamic communication strategy generation for multi-swarm particle swarm optimization framework (L2D-MSO). By constructing natural language prompts that encode the status information of each swarm, L2D-MSO enables real-time reasoning via large language models to dynamically generate communication strategy, guiding information exchange among multiple swarms to accelerate convergence and improve solution quality. In addition, an adaptive temperature mechanism is introduced to adjust the perturbation intensity according to the optimization progress, further enhancing the phase adaptability of strategy generation. Experimental results on the CEC2022 benchmark, robotic dog inspection path planning, and cloud-edge collaborative energy-aware scheduling tasks demonstrate the superior performance of L2D-MSO, validating its advantages in convergence accuracy, search stability, and high-dimensional adaptability across both physical and computational optimization scenarios.

Details DOI

EAAI Journal 2025 Journal Article

A surrogate-assist quasi-affine transformation evolutionary for multi-objective optimization of empty train deployment on heavy-haul railways

Zhi-Gang Du
Jeng-Shyang Pan
He-Ying Xu
Shu-Chuan Chu
Shao-Quan Ni

Surrogate-assisted evolutionary algorithms have become increasingly popular for solving expensive and time-consuming single-objective and multi-objective problems. However, in multi-objective optimization, the Pareto non-dominated solution set space can quickly become computationally intractable, making it challenging to select which samples to evaluate using the expensive fitness function. In this paper, we propose a Surrogate-Assist Quasi-Affine Transformation Evolutionary algorithm (SA-QUATRE/MO) for solving multi-objective optimization problems. The SA-QUATRE/MO algorithm uses a radial basis function network as a surrogate model to improve the speed of the algorithm operation by replacing the expensive fitness evaluation with the surrogate model. To ensure the excellence and diversity of the selected samples while keeping the archived sample space fixed, we propose a technique called Vector Space Sampling, which samples objective points in the current set of non-dominated solutions by dividing several sub-vector spaces. Additionally, we propose an uncertain sample infilling strategy to select samples for real fitness evaluation using a designed uncertainty function. We compare the SA-QUATRE/MO algorithm with three state-of-the-art algorithms for multi-objective problems in three test function suites. Finally, we applied the SA-QUATRE/MO algorithm to optimize empty train deployment in a heavy-haul railway at the loading end and build a model based on the S12 section of a specific railway. The final experimental results demonstrate the practicality and effectiveness of our proposed method.

Details DOI

JBHI Journal 2025 Journal Article

SleepHybridNet: A Lightweight Hybrid CNN-Transformer Model for Enhanced N1 Sleep Staging From Single-Channel EEG

Hao Zhou
Mengxiang Su
Jeng-Shyang Pan
Chenglong Dai
Ying Chen
Shu-Chuan Chu

This study introduces SleepHybridNet, a lightweight hybrid CNN-Transformer model designed to enhance the classification of non-rapid eye movement stage 1 (N1) sleep using single-channel electroencephalogram (EEG) signals. Accurate identification of the N1 stage is of critical importance in both sleep neuroscience and clinical practice. However, due to the ambiguous features during N1 stage, current deep learning models still struggle to achieve satisfactory performance. To address these challenges, SleepHybridNet integrates multi-scale feature fusion and sequence modeling through a novel architecture. It consists of a Multi-Scale Convolutional Neural Network (MSCNN) module, a Transformer encoder, a spectral feature extraction unit, and a multi-task classifier. Experimental results based on the publicly available Sleep-EDF Expanded dataset demonstrate that SleepHybridNet outperforms existing methods in both classification accuracy and generalization capability. Specifically, the model achieves an overall accuracy of 88. 2% and an F1-score of 0. 633 for the N1 stage, showing superior performance particularly in underrepresented classes such as N1 and N3 stages. With only 5. 1 M parameters, the lightweight design of the model can enable practical deployment in clinical settings, bridging the gap between high-performance deep learning algorithms and practical applicability in sleep medicine. Future work may explore the integration of multimodal data from wearable sensors to further expand its use in diverse application scenarios.

Details DOI

EAAI Journal 2024 Journal Article

Analysis on fetal phonocardiography segmentation problem by hybridized classifier

Lingping Kong
Katerina Barnova
Rene Jaros
Seyedali Mirjalili
Vaclav Snasel
Jeng-Shyang Pan
Radek Martinek

Fetal examinations are a significant and challenging field of healthcare. Cardiotocography is the most commonly used method for monitoring fetal heart rate and uterine contractions. As a promising alternative to cardiotocography, fetal phonocardiography is beginning to emerge. It is an entirely non-invasive, passive, and low-cost method. However, it is tough to estimate the ideal form of the fetal sound signal in most cases due to the presence of disturbances. The disturbances originate from movements or rotations of the fetal body, making fetal heart sound processing difficult. This study presents an automatic method for segmenting the fetal heart sounds in a phonocardiographic signal that is loaded with different types of disturbances and analyzes which of these disturbances most affect segmentation accuracy. To provide a comprehensive investigation, we propose a hybrid classifier based on Transformer and eXtreme Gradient Boosting, short for XGBoost, to improve segmentation performance by decision-making integration. 2000 segments of data from the Research Resource for Complex Physiologic Signals, PhysioNet repository, and created synthetic data (873 recordings) were used for the experiment. In the S1 label, our proposed method ranks first among all compared algorithms in precision, recall, F1, and accuracy score, tying with Transformer in recall score. It achieves an accuracy increase of 5% and 1. 3% compared to XGBoost and Transformer, respectively. Similarly, in the S2 label, there is a precision score increase of 5. 8% and 3. 7% compared to XGBoost and Transformer, respectively. In general, our proposed method shows effective and promising performance. .

Details DOI

EAAI Journal 2024 Journal Article

Flexible margins and multiple samples learning to enhance lexical semantic similarity

Jeng-Shyang Pan
Xiao Wang
Dongqiang Yang
Ning Li
Kevin Huang
Shu-Chuan Chu

The advancement of deep learning and neural networks has led to the widespread adoption of neural word embeddings as a prominent lexical representation method in natural language processing. With the help of the neural language model trained by the contextual information of large scale text, the neural word embedding obtained by the neural language model captures more semantic correlation in the semantic space, while ignoring the semantic similarity. It will incur high computational cost and time costs during the training process of the model. To better inject semantic similarity into the distribution space and reduce time cost, we perform post processing learning of neural word embeddings using deep metric learning. This paper proposes a lexical enhancement method based on flexible margins and multiple samples learning. In this method, we embed the lexical entailment constraint relations into neural word embeddings. By categorizing the set of lexical constraints and penalizing the negative samples to different degrees according to the gap between categories, and allowing the positive and negative samples to learn from each other in the distributed space. The method we propose significantly improves neural word embeddings. By evaluating neural word embedded vocabulary similarity, the benchmark accuracy is improved to 75%. The method shows great competitiveness in text similarity tasks and text categorization tasks. These findings summarize research results and provide strong support for further applications.

Details DOI

EAAI Journal 2022 Journal Article

A parallel compact firefly algorithm for the control of variable pitch wind turbine

Jie Shan
Shu-Chuan Chu
Shao-Wei Weng
Jeng-Shyang Pan
Shi-Jie Jiang
Shi-Guang Zheng

Firefly algorithm (FA) is a meta-heuristic optimization algorithm inspired by the flickering behavior of fireflies. Due to its excellent performance, it has been widely used in engineering field. However, FA is a population-based algorithm, which needs to occupy a lot of running memory. It is adverse for some small wind turbines or scenarios with limited memory Therefore, this paper proposes an improved firefly algorithm, called parallel compact firefly algorithm (PCFA). The compact technique helps FA save operation memory, which is advantageous for some usage scenarios limited memory. And the parallel technique helps compact FA achieve better solutions and faster convergence. The proposed PCFA was tested on 28 benchmark functions and applied to the proportional–integral–derivative (PID) parameter tuning of the variable pitch wind turbine. Results demonstrate that (1) PCFA is superior to common compact optimization algorithms not only on less memory consumption, but also on more competitive solutions and faster convergence. (2) PCFA shows the better applicability in the PID parameter tuning of variable pitch wind turbine. It can availably smooth the power output of wind turbine under less memory consumption.

Details DOI

EAAI Journal 2021 Journal Article

Digital watermarking with improved SMS applied for QR code

Jeng-Shyang Pan
Xiao-Xue Sun
Shu-Chuan Chu
Ajith Abraham
Bin Yan

With the rapid development of information technology, infringements have become increasingly serious. Digital watermarking is an effective method to protect information. The current watermarking technology still has room for further improvement in imperceptibility and robustness. This paper proposes an improved watermarking technology using meta-heuristic algorithm. Further, Quick Response code (QR code) is used as a carrier to transmit information. The improved Discrete Wavelet Transform-Singular Value Decomposition (DWT-SVD) is used to hide the watermark into the QR code. Therefore, digital watermarking is realized on the QR code. In the common watermark embedding methods, the digital watermark is related to the embedding strength. How to find a suitable embedding factor and reduce distortion is of great significance to these watermarking algorithms. This paper mainly proposes two novel algorithms based on States of Matter Search (SMS) algorithm to find suitable embedding factors. The first algorithm uses an adaptive parameter to control the movement of particles called the adaptive step States of Matter Search (sSMS). The second algorithm incorporates co-evolutionary matrix to enhance the search capability named Co-evolution States of Matter Search (CSMS). DWT-SVD is updated through two algorithms to acquire optimal embedding strength factors on the QR code watermarking. By adjusting the embedding strength factors, the intensity of the watermark embedded in different frequency domains would be modified. The experimental results have higher PSNR and the QR code can still be decoded by a general decoder. It shows that the proposed approaches are practicable and effective.

Details DOI

AAAI Conference 2016 Conference Paper

BRBA: A Blocking-Based Association Rule Hiding Method

Peng Cheng
Ivan Lee
Li Li
Kuo-Kun Tseng
Jeng-Shyang Pan

Privacy preserving in association rule mining is an important research topic in the database security ﬁeld. This paper has proposed a blocking-based method to solve the association rule hiding problem for data sharing. It aims at reducing undesirable side effects and increasing desirable side effects, while ensuring to conceal all sensitive rules. The candidate transactions are selected for sanitization based on their relations with border rules. Comparative experiments on real datasets demonstrate that the proposed method can achieve its goals.

PDF Details

JBHI Journal 2014 Journal Article

A Collaborative Computing Framework of Cloud Network and WBSN Applied to Fall Detection and 3-D Motion Reconstruction

Chin-Feng Lai
Min Chen
Jeng-Shyang Pan
Chan-Hyun Youn
Han-Chieh Chao

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.

Details DOI

AAAI Conference 2014 Conference Paper

Association Rule Hiding Based on Evolutionary Multi-Objective Optimization by Removing Items

Peng Cheng
Jeng-Shyang Pan

Today, people benefit from utilizing data mining technologies, such as association rule mining methods, to find valuable knowledge residing in a large amount of data. However, they also face the risk of exposing sensitive or confidential information, when data is shared among different organizations. Thus, a question arises: how can we prevent that sensitive knowledge is discovered, while ensuring that ordinary non-sensitive knowledge can be mined to the maximum extent possible. In this paper, we address the problem of privacy preserving in association rule mining from the perspective of multi-objective optimization. A new hiding method based on evolutionary multi-objective optimization (EMO) is proposed and the side effects generated by the hiding process are formulated as optimization goals. EMO is used to find candidate transactions to modify so that side effects are minimized. Comparative experiments with exact methods on real datasets demonstrated that the proposed method can hide sensitive rules with fewer side effects.

PDF Details