DLDonghun Lee

Donghun Lee

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Donghun Lee has a total of 10 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Quality-Controlled Generative Augmentation for North Atlantic Right Whale Upcall Detection Using Contour 1D-VAE

    by Jongmin Ahn, Geun-Ho Park, Ho-Seuk Bae, Donghun Lee

    This study proposes a Variational AutoEncoder (VAE)-based Quality Controlled (QC) generative augmentation framework for North Atlantic right whale (NARW) upcall detection. Existing generative augmentation methods can generate synthetic samples; however, they do not provide a sample-level criterion for determining whether each generated sample is a positive sample that contributes to improved detector performance or a synthetic outlier that should be removed. This study learns manually extracted upcall frequency contours using a 1D-VAE and evaluates generated contour candidates in a 10-dimensional acoustic morphology feature space. The QC score is computed with respect to the reference distribution of manually extracted real upcall contours, and stochastic acceptance probabilities for borderline samples around the hard threshold are calibrated using same-call manual re-extraction variability. QC-passed contours are converted into detector training spectrograms using smooth amplitude modulation and Gaussian noise injection based on SNR statistics. Using 30,000 acoustic segments from the Kaggle NARW dataset, Original, Denoising Diffusion Probabilistic Models (DDPM), Contour-VAE without QC, and VAE-QC conditions were compared under the same detector and augmentation budget. VAE-QC with α = 0.97 achieved the highest mean AUC of 0.901, outperforming Original training (0.802), DDPM (0.845), and Contour-VAE without QC (0.810). Feature distribution and QC-score analyses further showed that VAE-QC suppresses morphology outlier tails observed in unfiltered generation. These results indicate that the key factor in generative augmentation is the QC process that defines feature boundaries useful for detector learning and selects synthetic positive samples accordingly.

  • Open Access

    Design Procedure of Cascaded Multilevel Inverter for High-Power Amplifier in SONAR System

    by Jejin Jang, Jaehyuk Choi, Donghun Lee, Hyungsoo Mok

    In recent years, there has been a trend toward expanding the operating frequency range and increasing the output power of Sound Navigation and Ranging (SONAR) systems to enhance their acoustic detection capabilities. However, due to this increase in operational power, the electrical capacity of amplifiers for SONAR system operation also increases, necessitating High-Power Amplifiers. When configured with a single amplifier, as in conventional methods, the volume of amplifiers increases due to volumetric increases in heat dissipation, components, and windings. These issues are detrimental to SONAR amplifier installation, mobility, maintenance, and equipment lifespan due to stress on individual components. Additionally, amplifiers for SONAR systems are comprised of power conversion devices, transformers for LC filters and matching, necessitating consideration of LC filters and matching transformers for enhancing voltage quality and efficiency to improve amplifier performance transmitted to SONAR transducers. However, previous research has focused on single-amplifier design methods, neglecting such considerations. Therefore, this paper proposes a design technique that overcomes the drawbacks of using the conventional design method by configuring multiple H-bridge inverters in a cascade format and utilizes one of the optimization algorithms, Particle Swarm Optimization (PSO), to derive amplifier design techniques that optimize component parameters for enhancing high-capacity amplifier performance. Subsequently, theoretical analysis, simulations, and experimental results comparing the proposed high-power amplifier design method with conventional single-amplifier design methods demonstrate similar error rates in operational frequency bands.


  • Open Access

    Estimation Method of an Electrical Equivalent Circuit for Sonar Transducer Impedance Characteristic of Multiple Resonance

    by Jejin Jang, Jaehyuk Choi, Donghun Lee, Hyungsoo Mok

    Improving the operational efficiency and optimizing the design of sound navigation and ranging (sonar) systems require accurate electrical equivalent models within the operating frequency range. The power conversion system within the sonar system increases power efficiency through impedance-matching circuits. Impedance matching is used to enhance the power transmission efficiency of the sonar system. Therefore, to increase the efficiency of the sonar system, an electrical-matching circuit is employed, and this necessitates an accurate equivalent circuit for the sonar transducer within the operating frequency range. In conventional equivalent circuit derivation methods, errors occur because they utilize the same number of RLC branches as the resonant frequency of the sonar transducer, based on its physical properties. Hence, this paper proposes an algorithm for deriving an equivalent circuit independent of resonance by employing multiple electrical components and particle swarm optimization (PSO). A comparative verification was also performed between the proposed and existing approaches using the Butterworth–van Dyke (BVD) model, which is a method for deriving electrical equivalent circuits.