Institution profile

National Chung Cheng University

Academic institutionasia · tw
Official website
Research library19linked papers
Opportunities0open roles
Selected work

Representative Papers

Spatiotemporal Continual Learning for Mobile Edge UAV Networks: Mitigating Catastrophic Forgetting

Jan 29, 2026

This work addresses the challenge of catastrophic forgetting in conventional deep reinforcement learning approaches when mobile edge drone networks undergo abrupt user distribution shifts during dynamic spatiotemporal scenario transitions, such as from urban to rural environments, which often necessitates frequent retraining and causes service interruptions. To mitigate this, the authors propose a Spatiotemporal Continual Learning (STCL) framework that integrates Group-Decoupled Multi-Agent Proximal Policy Optimization (G-MAPPO) with dynamic z-score normalization. The framework employs a Group-Decoupled Policy Optimization (GDPO) mechanism to online balance heterogeneous objectives—including energy efficiency, fairness, and coverage—and leverages 3D drone mobility as a spatial compensation layer. Experimental results demonstrate that the proposed method restores service reliability to approximately 0.95 after scenario transitions and achieves a 20% higher effective capacity than MADDPG under extreme load, significantly alleviating knowledge forgetting while ensuring service continuity.

1 citationsRead paper

Predictive Lightweight MARL for Resilient Coverage in Sparse-Signaling Aerial Networks

Jul 24, 2026

This work addresses the challenge of coordination collapse in bandwidth-constrained drone swarms caused by sparse communication and information staleness. To this end, we propose a Predictive Lightweight Multi-Agent Reinforcement Learning framework (PL-MARL), which innovatively integrates a kinematics-aware active inference mechanism into a lightweight MARL architecture. By leveraging physical priors to proactively reconstruct neighboring agents’ trajectories, PL-MARL achieves an efficient trade-off between computation and communication under extremely low bandwidth overhead, effectively decoupling system structural resilience from communication frequency. Experimental results demonstrate that PL-MARL maintains high coverage performance and task continuity even under extreme communication scarcity and node failures, significantly enhancing robustness against disturbances while conserving spectral resources for payload operations.

0 citationsRead paper

Algebraic Machine Learning for Small-to-Medium Datasets Is Competitive against Strong Standard Baselines

May 21, 2026

This work proposes a machine learning approach grounded in the subdirect decomposition of algebraic structures, introducing a universal algebraic inductive bias that requires no modality-specific design, task-dependent hyperparameters, cross-validation, or numerical optimization. On small to medium-sized datasets (50–2000 samples), where existing symbolic methods often struggle to match modern strong baselines, the proposed method demonstrates competitive performance: it outperforms cross-validated convolutional neural networks (CNNs) on image classification tasks and achieves results comparable to LightGBM and random forests on tabular data. These findings highlight the potential of algebraic machine learning to generalize effectively in low-data regimes without relying on conventional optimization or extensive hyperparameter tuning.

0 citationsRead paper
Recent publications

Latest Papers

Predictive Lightweight MARL for Resilient Coverage in Sparse-Signaling Aerial Networks

Jul 24, 2026

This work addresses the challenge of coordination collapse in bandwidth-constrained drone swarms caused by sparse communication and information staleness. To this end, we propose a Predictive Lightweight Multi-Agent Reinforcement Learning framework (PL-MARL), which innovatively integrates a kinematics-aware active inference mechanism into a lightweight MARL architecture. By leveraging physical priors to proactively reconstruct neighboring agents’ trajectories, PL-MARL achieves an efficient trade-off between computation and communication under extremely low bandwidth overhead, effectively decoupling system structural resilience from communication frequency. Experimental results demonstrate that PL-MARL maintains high coverage performance and task continuity even under extreme communication scarcity and node failures, significantly enhancing robustness against disturbances while conserving spectral resources for payload operations.

0 citationsRead paper

Algebraic Machine Learning for Small-to-Medium Datasets Is Competitive against Strong Standard Baselines

May 21, 2026

This work proposes a machine learning approach grounded in the subdirect decomposition of algebraic structures, introducing a universal algebraic inductive bias that requires no modality-specific design, task-dependent hyperparameters, cross-validation, or numerical optimization. On small to medium-sized datasets (50–2000 samples), where existing symbolic methods often struggle to match modern strong baselines, the proposed method demonstrates competitive performance: it outperforms cross-validated convolutional neural networks (CNNs) on image classification tasks and achieves results comparable to LightGBM and random forests on tabular data. These findings highlight the potential of algebraic machine learning to generalize effectively in low-data regimes without relying on conventional optimization or extensive hyperparameter tuning.

0 citationsRead paper

Teachers' Vocal Expressions and Student Engagement in Asynchronous Video Learning

May 17, 2026

This study addresses the issue of insufficient emotional engagement among students in asynchronous video-based learning by systematically distinguishing and comparing the differential effects of instructors’ verbal versus nonverbal vocal emotions. Drawing on data from 210 MOOC videos and feedback from 738 learners, the research integrates computational acoustic analysis, textual sentiment analysis, and classification of six basic nonverbal vocal emotion categories. Findings reveal that high-arousal positive nonverbal emotions—such as happiness and surprise—significantly enhance learners’ emotional engagement, whereas high-arousal negative emotions like anger markedly diminish it. In contrast, verbal emotional content shows no significant impact on engagement. These results underscore the critical role of instructors’ nonverbal vocal expressions in fostering emotional engagement in online learning environments.

0 citationsRead paper