arXiv subjects

Sangwoo Kang

Publications and source records attributed to Sangwoo Kang.

12 recordsLinked to original sources

arXiv2026

KLOD: Locality-Preserving Knowledge Editing via Non-Target Distribution Preservation

Fine-tuning-based knowledge editing is simple and architecture-agnostic, but standard cross-entropy increases the edited target probability without explicitly constraining changes in the non-target output distribution. In sequential editing, such unconstrained redistribution can accumulate as distributional drift and contribute to locality degradation. We propose KLOD, a bounded and distribution-preserving objective for fine-tuning-based knowledge editing that separates the intended target update from distributions that should remain stable. KLOD stops target amplification once a probability threshold is reached, while preserving the target-excluded non-target distribution at target positions and the full next-token distribution at prefix positions. Experiments on CounterFact and ZsRE with Llama3-8B-Instruct and Qwen2.5-7B-Instruct show that KLOD substantially mitigates locality degradation while maintaining high edit reliability. The target probability threshold further provides a controllable Generalization--Locality trade-off. Ablation, multi-seed, and distributional KL analyses support the interpretation that KLOD's locality gains are associated with preserving output distributions rather than simply weakening the edit. Code is available on GitHub https://github.com/Hostoday/KLOD .

cs.AI↗

arXiv2026

Bayesian optimization approach for tracking a moving target from far-field data in three dimensions

We investigate a three-dimensional inverse scattering problem for tracking a rigidly moving target from far-field data generated by a single incident field. Extending our recent two-dimensional study, we develop a Bayesian optimization framework for simultaneously tracking the target's location and orientation over successive time steps, with the translational and rotational motions modeled as independent stochastic processes. We derive analytical formulas for the far-field pattern under translations and rotations and use them to design a Bayesian optimization procedure tailored to the tracking problem. We further establish posterior consistency for the underlying probabilistic model. When the target shape is unknown, its shape is identified at the initial time using a fully connected neural network trained on a precomputed dataset. Numerical experiments validate the effectiveness of the proposed framework.

math.AP↗

arXiv2025

Real-time inversion of two-dimensional Fresnel experimental database using orthogonality sampling method with single and multiple sources: the case of transverse electric polarized waves

This paper concerns an application of the orthogonality sampling method (OSM) for a real-time identification of small objects from two-dimensional Fresnel experimental dataset in transverse electric polarization. First, we apply the OSM with a single source by designing an indicator function based on the asymptotic expansion formula for the scattered field in the presence of small objects. We demonstrate that the indicator function can be expressed by an infinite series of Bessel functions of integer order of the first kind, the range of the signal receiver, and the location of the emitter. Based on this, we then investigate the applicability and limitations of the designed OSM. Specifically, we find that the imaging performance is strongly dependent on the source and the applied frequency. We then apply the OSM with multiple sources to improve imaging performance. Based on the identified structure of the OSM with a single source, we design an indicator function with multiple sources and demonstrate that it can be expressed by an infinite series of the Bessel function of integer order of the first kind, and we explain that objects can be identified uniquely using the designed OSM. Numerical simulation results obtained with the Fresnel experimental dataset demonstrate the advantages and disadvantages of the OSM with a single source and confirm that the designed OSM with multiple sources improves imaging performance.

math.NA↗

arXiv2024

Bayesian optimization approach for tracking the location and orientation of a moving target using far-field data

We investigate the inverse scattering problem for tracking the location and orientation of a moving scatterer using a single incident field. We solve the problem by adopting the optimization approach with the objective function defined by the discrepancy in far-field data. We rigorously derive formulas for the far-field data under translation and rotation of the target and prove that the objective function is locally Lipschitz with respect to the orientation angle at the true angle. By integrating these formulas with the Bayesian optimization approach, we reduce the cost of objective function evaluations. For the instance of an unknown target, machine learning via fully connected neural networks is applied to identify the shape of the target. Numerical simulations for randomly generated shapes and trajectories demonstrate the effectiveness of the proposed method.

math.AP↗

arXiv2023

Monostatic imaging of an extended target with MCMC sampling

We consider the imaging of a planar extended target from far-field data under a monostatic measurement configuration, in which the data is measured by a single moving transducer, as frequently encountered in practical application. In this paper, we develop a Bayesian approach to recover the shape of the extended target with MCMC sampling, where a new shape basis selection is proposed based on the shape derivative analysis for the measurement data. In order to optimize the center and radius of the initial disk, we use the monostatic sampling method for the center and the explicit scattered field expression for disks for the radius. Numerical simulations are presented to validate the proposed method.

math.NA↗

arXiv2023

Inverse problem for a planar conductivity inclusion

This paper concerns the inverse problem of determining a planar conductivity inclusion. Our aim is to analytically recover from the generalized polarization tensors (GPTs), which can be obtained from exterior measurements, a homogeneous inclusion with arbitrary constant conductivity. The primary outcome of recovering a homogeneous inclusion is an inversion formula in terms of the GPTs for conformal mapping coefficients associated with the inclusion. To prove the formula, we establish matrix factorizations for the GPTs.

math.AP↗

arXiv2022

Monostatic sampling methods in limited-aperture configuration

We present monostatic sampling methods for limited-aperture scattering problems in two dimensions. The direct sampling method (DSM) is well known to provide a robust, stable, and fast numerical scheme for imaging inhomogeneities from multistatic measurements even with only one or two incident fields. However, in practical applications, monostatic measurements in limited-aperture configuration are frequently encountered. A monostatic sampling method (MSM) was studied in full-aperture configuration in recent literature. In this paper, we develop MSM in limited-aperture configuration and derive an asymptotic formula of the corresponding indicator function. Based on the asymptotic formula, we then analyze the imaging performance of the proposed method depending on the range of measurement directions and the geometric, material properties of inhomogeneities. Furthermore, we propose a modified numerical scheme with multi-frequency measurements that improve imaging performance, especially for small anomalies. Numerical simulations are presented to validate the analytical results.

math.NA↗

arXiv2019

Analysis and improvement of direct sampling method in the mono-static configuration

The recently introduced non-iterative imaging method entitled \enquote{direct sampling method} (DSM) is known to be fast, robust, and effective for inverse scattering problems in the multi-static configuration but fails when applied to the mono-static one. To the best of our knowledge no explanation of this failure has been provided yet. Thanks to the framework of the asymptotic and the far-field hypothesis in the 2D scalar configuration an analytical expression of the DSM indicator function in terms of the Bessel function of order zero and sizes, shapes and permittivities of the inhomogeneities is obtained and the theoretical reason of the limitation identified. A modified version of DSM is then proposed in order to improve the imaging method. The theoretical results are supported by numerical results using synthetic data.

math.NA↗

arXiv2018

Direct sampling method for imaging small dielectric inhomogeneities: analysis and improvement

The direct sampling method (DSM) has been introduced for non-iterative imaging of small inhomogeneities and is known to be fast, robust, and effective for inverse scattering problems. However, to the best of our knowledge, a full analysis of the behavior of the DSM has not been provided yet. Such an analysis is proposed here within the framework of the asymptotic hypothesis in the 2D case leading to the expression of the DSM indicator function in terms of the Bessel function of order zero and the sizes, shapes and permittivities of the inhomogeneities. Thanks to this analytical expression the limitations of the DSM method when one of the inhomogeneities is smaller and/or has lower permittivity than the others is exhibited and illustrated. An improved DSM is proposed to overcome this intrinsic limitation in the case of multiple incident waves. Then we show that both the traditional and improved DSM are closely related to a normalized version of the Kirchhoff migration. The theoretical elements of our proposal are supported by various results from numerical simulations with synthetic and experimental data.

math.NA↗

arXiv2016

Air Stable Doping and Intrinsic Mobility Enhancement in Monolayer $MoS_{2}$ by Amorphous $TiO_{x}$ Encapsulation

To reduce Schottky-barrier-induced contact and access resistance, and the impact of charged impurity and phonon scattering on mobility in devices based on 2D transition metal dichalcogenides (TMDs), considerable effort has been put into exploring various doping techniques and dielectric engineering using $high-κ$ oxides, respectively. The goal of this work is to demonstrate a $high-κ$ dielectric that serves as an effective n-type charge transfer dopant on monolayer (ML) molybdenum disulfide ($MoS_{2}$). Utilizing amorphous titanium suboxide (ATO) as the '$high-κ$ dopant', we achieved a contact resistance of ~ $180$ $Ω.μm$ which is the lowest reported value for ML $MoS_{2}$. An ON current as high as $240$ $μA/μm$ and field effect mobility as high as $83$ $cm^2/V-s$ were realized using this doping technique. Moreover, intrinsic mobility as high as $102$ $cm^2/V-s$ at $300$ $K$ and $501$ $cm^2/V-s$ at $77$ $K$ were achieved after ATO encapsulation which are among the highest mobility values reported on ML $MoS_{2}$. We also analyzed the doping effect of ATO films on ML $MoS_{2}$, a phenomenon which is absent when stoichiometric $TiO_{2}$ is used, using ab initio density functional theory (DFT) calculations which shows excellent agreement with our experimental findings. Based on the interfacial-oxygen-vacancy mediated doping as seen in the case of $high-κ$ ATO - ML $MoS_{2}$, we propose a mechanism for the mobility enhancement effect observed in TMD-based devices after encapsulation in a $high-κ$ dielectric environment.

cond-mat.mes-hall↗

arXiv2015

High-Mobility Holes in Dual-Gated WSe$_2$ Field-Effect Transistors

We demonstrate dual-gated $p$-type field-effect transistors (FETs) based on few-layer tungsten diselenide (WSe$_2$) using high work-function platinum source/drain contacts, and a hexagonal boron nitride top-gate dielectric. A device topology with contacts underneath the WSe$_2$ results in $p$-FETs with $I_{ON}$/$I_{OFF}$ ratios exceeding 10$^7$, and contacts that remain Ohmic down to cryogenic temperatures. The output characteristics show current saturation and gate tunable negative differential resistance. The devices show intrinsic hole mobilities around 140 cm$^2$/Vs at room temperature, and approaching 4,000 cm$^2$/Vs at 2 K. Temperature-dependent transport measurements show a metal-insulator transition, with an insulating phase at low densities, and a metallic phase at high densities. The mobility shows a strong temperature dependence consistent with phonon scattering, and saturates at low temperatures, possibly limited by Coulomb scattering, or defects.

cond-mat.mes-hall↗

arXiv2014

Gate-Tunable Resonant Tunneling in Double Bilayer Graphene Heterostructures

We demonstrate gate-tunable resonant tunneling and negative differential resistance in the interlayer current-voltage characteristics of rotationally aligned double bilayer graphene heterostructures separated by hexagonal boron-nitride (hBN) dielectric. An analysis of the heterostructure band alignment using individual layer densities, along with experimentally determined layer chemical potentials indicates that the resonance occurs when the energy bands of the two bilayer graphene are aligned. We discuss the tunneling resistance dependence on the interlayer hBN thickness, as well as the resonance width dependence on mobility and rotational alignment.

cond-mat.mes-hall↗