Assembling the CREW: A Collaborative Multi-agent Reinforcement Learning Framework for Automated Related Work Generation
本文提出CREW框架，通过多智能体强化学习动态协作生成相关工作综述，解决了传统方法中静态协作的局限性，提高了文献合成的质量并降低了成本。
0 citationsRead paper
Selected work
Representative Papers
Comparing Methodologies for Ranking Alternatives: A case study in assessing bank financial performance
Addressing the challenge of cross-institutional financial performance comparison arising from indicator heterogeneity in banking, this study systematically evaluates the applicability of multi-criteria decision-making (MCDM) methods for interbank performance assessment. Innovatively, it conducts the first comprehensive empirical analysis integrating five objective weighting methods—Entropy Weight, MEREC, LOPCOW, SPC, and CRITIC—with three ranking techniques—Probabilistic Ranking, TOPSIS, and RAM—using data from 19 commercial banks. Results demonstrate that the Entropy Weight–Probabilistic Ranking combination significantly outperforms all other methodological pairings in ranking stability, consistency, and robustness to perturbations. This provides a more reliable and reproducible quantitative framework for bank performance benchmarking. The study fills a critical gap by delivering the first systematic, evidence-based validation of MCDM methodologies in financial supervision and peer-group benchmarking contexts.
0 citationsRead paper
A Comparative Analysis of Multi-Criteria Decision-Making (MCDM) Methods
This study addresses the inconsistency of multi-criteria decision-making (MCDM) methods in bank performance evaluation. It systematically compares the ranking outcomes of four MCDM methods—RAM, MOORA, FUCA, and CURLI—applied to 30 Vietnamese commercial banks, benchmarking results against the authoritative CAMELS rating framework. For the first time, FUCA and CURLI are employed for bank ranking, utilizing a six-dimensional indicator system: capital adequacy, asset quality, management capability, profitability, liquidity, and market risk sensitivity. Method effectiveness is quantified via Spearman’s rank correlation coefficient (ρ). Results show FUCA and CURLI achieve exceptional alignment with CAMELS (ρ = 0.9996 and 0.9984, respectively), whereas RAM and MOORA exhibit negative correlations, indicating poor suitability. This work not only clarifies the differential applicability of MCDM methods in financial supervision contexts but also extends the empirical application frontier of FUCA and CURLI in financial performance assessment.
0 citationsRead paper
Robust adaptive fuzzy sliding mode control for trajectory tracking for of cylindrical manipulator
Cylindrical robotic manipulators employed in high-precision applications—such as CNC machining and 3D printing—suffer from degraded trajectory tracking accuracy, severe chattering, and insufficient robustness due to system uncertainties and external disturbances. To address these challenges, this paper proposes a robust adaptive fuzzy sliding mode control (RAFSMC) scheme. The method integrates a fuzzy logic system for online approximation of unknown dynamics, an adaptive law for real-time parameter updating, and sliding mode control to ensure steady-state performance; closed-loop system stability is rigorously proven via Lyapunov theory. Compared with conventional sliding mode or PID controllers, the proposed approach significantly suppresses chattering, improves trajectory tracking accuracy—by approximately 42% in simulation—and enhances disturbance rejection capability. Moreover, it exhibits superior parametric adaptability and engineering practicality, offering a verifiable control solution for precision industrial robotics.
0 citationsRead paper
Optimization of sliding control parameters for a 3-dof robot arm using genetic algorithm (GA)
To address low trajectory tracking accuracy and severe chattering in conventional sliding mode control (SMC) for a 3-DOF robotic manipulator under model uncertainties and external disturbances, this paper proposes a robust control strategy that employs a genetic algorithm (GA) to optimize key SMC parameters. Leveraging its global search capability, the GA automatically tunes critical parameters—including the switching gain and boundary layer thickness—thereby preserving strong robustness while significantly mitigating chattering. Unlike traditional and fuzzy SMC approaches, the proposed method requires no prior knowledge and enables adaptive parameter adjustment. Simulation results demonstrate substantial improvements: trajectory tracking error is reduced by approximately 42%, and control chattering is markedly suppressed. The approach thus achieves a favorable balance among high tracking precision, strong robustness against uncertainties and disturbances, and practical implementability in engineering applications.
0 citationsRead paper
Recent publications
Latest Papers
Assembling the CREW: A Collaborative Multi-agent Reinforcement Learning Framework for Automated Related Work Generation
本文提出CREW框架，通过多智能体强化学习动态协作生成相关工作综述，解决了传统方法中静态协作的局限性，提高了文献合成的质量并降低了成本。
0 citationsRead paper
Comparing Methodologies for Ranking Alternatives: A case study in assessing bank financial performance
Addressing the challenge of cross-institutional financial performance comparison arising from indicator heterogeneity in banking, this study systematically evaluates the applicability of multi-criteria decision-making (MCDM) methods for interbank performance assessment. Innovatively, it conducts the first comprehensive empirical analysis integrating five objective weighting methods—Entropy Weight, MEREC, LOPCOW, SPC, and CRITIC—with three ranking techniques—Probabilistic Ranking, TOPSIS, and RAM—using data from 19 commercial banks. Results demonstrate that the Entropy Weight–Probabilistic Ranking combination significantly outperforms all other methodological pairings in ranking stability, consistency, and robustness to perturbations. This provides a more reliable and reproducible quantitative framework for bank performance benchmarking. The study fills a critical gap by delivering the first systematic, evidence-based validation of MCDM methodologies in financial supervision and peer-group benchmarking contexts.
0 citationsRead paper
A Comparative Analysis of Multi-Criteria Decision-Making (MCDM) Methods
This study addresses the inconsistency of multi-criteria decision-making (MCDM) methods in bank performance evaluation. It systematically compares the ranking outcomes of four MCDM methods—RAM, MOORA, FUCA, and CURLI—applied to 30 Vietnamese commercial banks, benchmarking results against the authoritative CAMELS rating framework. For the first time, FUCA and CURLI are employed for bank ranking, utilizing a six-dimensional indicator system: capital adequacy, asset quality, management capability, profitability, liquidity, and market risk sensitivity. Method effectiveness is quantified via Spearman’s rank correlation coefficient (ρ). Results show FUCA and CURLI achieve exceptional alignment with CAMELS (ρ = 0.9996 and 0.9984, respectively), whereas RAM and MOORA exhibit negative correlations, indicating poor suitability. This work not only clarifies the differential applicability of MCDM methods in financial supervision contexts but also extends the empirical application frontier of FUCA and CURLI in financial performance assessment.
0 citationsRead paper
Robust adaptive fuzzy sliding mode control for trajectory tracking for of cylindrical manipulator
Cylindrical robotic manipulators employed in high-precision applications—such as CNC machining and 3D printing—suffer from degraded trajectory tracking accuracy, severe chattering, and insufficient robustness due to system uncertainties and external disturbances. To address these challenges, this paper proposes a robust adaptive fuzzy sliding mode control (RAFSMC) scheme. The method integrates a fuzzy logic system for online approximation of unknown dynamics, an adaptive law for real-time parameter updating, and sliding mode control to ensure steady-state performance; closed-loop system stability is rigorously proven via Lyapunov theory. Compared with conventional sliding mode or PID controllers, the proposed approach significantly suppresses chattering, improves trajectory tracking accuracy—by approximately 42% in simulation—and enhances disturbance rejection capability. Moreover, it exhibits superior parametric adaptability and engineering practicality, offering a verifiable control solution for precision industrial robotics.
0 citationsRead paper
Optimization of sliding control parameters for a 3-dof robot arm using genetic algorithm (GA)
To address low trajectory tracking accuracy and severe chattering in conventional sliding mode control (SMC) for a 3-DOF robotic manipulator under model uncertainties and external disturbances, this paper proposes a robust control strategy that employs a genetic algorithm (GA) to optimize key SMC parameters. Leveraging its global search capability, the GA automatically tunes critical parameters—including the switching gain and boundary layer thickness—thereby preserving strong robustness while significantly mitigating chattering. Unlike traditional and fuzzy SMC approaches, the proposed method requires no prior knowledge and enables adaptive parameter adjustment. Simulation results demonstrate substantial improvements: trajectory tracking error is reduced by approximately 42%, and control chattering is markedly suppressed. The approach thus achieves a favorable balance among high tracking precision, strong robustness against uncertainties and disturbances, and practical implementability in engineering applications.
0 citationsRead paper