arXiv subjects

Haejun Chung

Publications and source records attributed to Haejun Chung.

At least 37 records · Page 2Linked to original sources

arXiv2026

Dodgersort: Uncertainty-Aware VLM-Guided Human-in-the-Loop Pairwise Ranking

Pairwise comparison labeling is emerging as it yields higher inter-rater reliability than conventional classification labeling, but exhaustive comparisons require quadratic cost. We propose Dodgersort, which leverages CLIP-based hierarchical pre-ordering, a neural ranking head and probabilistic ensemble (Elo, BTL, GP), epistemic--aleatoric uncertainty decomposition, and information-theoretic pair selection. It reduces human comparisons while improving the reliability of the rankings. In visual ranking tasks in medical imaging, historical dating, and aesthetics, Dodgersort achieves a 11--16\% annotation reduction while improving inter-rater reliability. Cross-domain ablations across four datasets show that neural adaptation and ensemble uncertainty are key to this gain. In FG-NET with ground-truth ages, the framework extracts 5--20$\times$ more ranking information per comparison than baselines, yielding Pareto-optimal accuracy--efficiency trade-offs.

cs.CV↗

arXiv2025

EZ-Sort: Efficient Pairwise Comparison via Zero-Shot CLIP-Based Pre-Ordering and Human-in-the-Loop Sorting

Pairwise comparison is often favored over absolute rating or ordinal classification in subjective or difficult annotation tasks due to its improved reliability. However, exhaustive comparisons require a massive number of annotations (O(n^2)). Recent work has greatly reduced the annotation burden (O(n log n)) by actively sampling pairwise comparisons using a sorting algorithm. We further improve annotation efficiency by (1) roughly pre-ordering items using the Contrastive Language-Image Pre-training (CLIP) model hierarchically without training, and (2) replacing easy, obvious human comparisons with automated comparisons. The proposed EZ-Sort first produces a CLIP-based zero-shot pre-ordering, then initializes bucket-aware Elo scores, and finally runs an uncertainty-guided human-in-the-loop MergeSort. Validation was conducted using various datasets: face-age estimation (FGNET), historical image chronology (DHCI), and retinal image quality assessment (EyePACS). It showed that EZ-Sort reduced human annotation cost by 90.5% compared to exhaustive pairwise comparisons and by 19.8% compared to prior work (when n = 100), while improving or maintaining inter-rater reliability. These results demonstrate that combining CLIP-based priors with uncertainty-aware sampling yields an efficient and scalable solution for pairwise ranking.

cs.CV↗

arXiv2025

Generalizing Deep Surrogate Solvers for Broadband Electromagnetic Field Prediction at Unseen Wavelengths

Recently, electromagnetic surrogate solvers, trained on solutions of Maxwell's equations under specific simulation conditions, enabled fast inference of computationally expensive simulations. However, conventional electromagnetic surrogate solvers often consider only a narrow range of spectrum and fail when encountering even slight variations in simulation conditions. To address this limitation, we define spectral consistency as the property by which the spatial frequency structure of wavelength-dependent condition embeddings matches that of the target electromagnetic field patterns. In addition, we propose two complementary components: a refined wave prior, which is the condition embedding that satisfies spectral consistency, and Wave-Informed element-wise Multiplicative Encoding (WIME), which integrates these embeddings throughout the model while preserving spectral consistency. This framework enables accurate field prediction across the broadband spectrum, including untrained intermediate wavelengths. Our approach reduces the normalized mean squared error at untrained wavelengths by up to 71% compared to the state-of-the-art electromagnetic surrogate solver and achieves a speedup of over 42 times relative to conventional numerical simulations.

cs.LG↗

arXiv2025

Inverse-Designed Metasurfaces for Wavefront Restoration in Under-Display Camera Systems

Under-display camera (UDC) systems enable full-screen displays in smartphones by embedding the camera beneath the display panel, eliminating the need for notches or punch holes. However, the periodic pixel structures of display panels introduce significant optical diffraction effects, leading to imaging artifacts and degraded visual quality. Conventional approaches to mitigate these distortions, such as deep learning-based image reconstruction, are often computationally expensive and unsuitable for real-time applications in consumer electronics. This work introduces an inverse-designed metasurface for wavefront restoration, addressing diffraction-induced distortions without relying on external software processing. The proposed metasurface effectively suppresses higher-order diffraction modes caused by the metallic pixel structures, restores the optical wavefront, and enhances imaging quality across multiple wavelengths. By eliminating the need for software-based post-processing, our approach establishes a scalable, real-time optical solution for diffraction management in UDC systems. This advancement paves the way to achieve software-free real-time image restoration frameworks for many industrial applications.

physics.optics↗

arXiv2025

Physics-guided and fabrication-aware inverse design of photonic devices using diffusion models

Designing free-form photonic devices is fundamentally challenging due to the vast number of possible geometries and the complex requirements of fabrication constraints. Traditional inverse-design approaches--whether driven by human intuition, global optimization, or adjoint-based gradient methods--often involve intricate binarization and filtering steps, while recent deep learning strategies demand prohibitively large numbers of simulations (10^5 to 10^6). To overcome these limitations, we present AdjointDiffusion, a physics-guided framework that integrates adjoint sensitivity gradients into the sampling process of diffusion models. AdjointDiffusion begins by training a diffusion network on a synthetic, fabrication-aware dataset of binary masks. During inference, we compute the adjoint gradient of a candidate structure and inject this physics-based guidance at each denoising step, steering the generative process toward high figure-of-merit (FoM) solutions without additional post-processing. We demonstrate our method on two canonical photonic design problems--a bent waveguide and a CMOS image sensor color router--and show that our method consistently outperforms state-of-the-art nonlinear optimizers (such as MMA and SLSQP) in both efficiency and manufacturability, while using orders of magnitude fewer simulations (approximately 2 x 10^2) than pure deep learning approaches (approximately 10^5 to 10^6). By eliminating complex binarization schedules and minimizing simulation overhead, AdjointDiffusion offers a streamlined, simulation-efficient, and fabrication-aware pipeline for next-generation photonic device design. Our open-source implementation is available at https://github.com/dongjin-seo2020/AdjointDiffusion.

physics.optics↗

arXiv2025

Inverse design of ultrathin metamaterial absorber

Electromagnetic absorbers combining ultrathin profiles with robust absorptivity across wide incidence angles are essential for applications such as stealth technology, wireless communications, and quantum computing. Traditional designs, including Salisbury screens, typically require thicknesses of at least a quarter-wavelength (lambda/4), which limits their use in compact systems. While metamaterial absorbers (MMAs) can reduce thickness, their absorptivity generally decreases under oblique incidence conditions. Here, we introduce an adjoint optimization-based inverse design method that merges the ultrathin advantage of MMAs with the angle-insensitive characteristics of Salisbury screens. By leveraging the computational efficiency of the adjoint method, we systematically optimize absorber structures as thin as lambda/20. The optimized designs achieve absorption exceeding 90% at the target frequency of 7.5 GHz and demonstrate robust performance under oblique incidence, maintaining over 90% absorption up to 50°, approximately 80% at 60°, and around 70% at 70°. Comparative analysis against particle swarm optimization highlights the superior efficiency of the adjoint method, reducing computational effort by approximately 98%. This inverse design framework thus provides substantial improvements in both performance and computational cost, offering a promising approach for advanced electromagnetic absorber design.

physics.optics↗

arXiv2025

BOLDSimNet: Examining Brain Network Similarity between Task and Resting-State fMRI

Traditional causal connectivity methods in task-based and resting-state functional magnetic resonance imaging (fMRI) face challenges in accurately capturing directed information flow due to their sensitivity to noise and inability to model multivariate dependencies. These limitations hinder the effective comparison of brain networks between cognitive states, making it difficult to analyze network reconfiguration during task and resting states. To address these issues, we propose BOLDSimNet, a novel framework utilizing Multivariate Transfer Entropy (MTE) to measure causal connectivity and network similarity across different cognitive states. Our method groups functionally similar regions of interest (ROIs) rather than spatially adjacent nodes, improving accuracy in network alignment. We applied BOLDSimNet to fMRI data from 40 healthy controls and found that children exhibited higher similarity scores between task and resting states compared to adolescents, indicating reduced variability in attention shifts. In contrast, adolescents showed more differences between task and resting states in the Dorsal Attention Network (DAN) and the Default Mode Network (DMN), reflecting enhanced network adaptability. These findings emphasize developmental variations in the reconfiguration of the causal brain network, showcasing BOLDSimNet's ability to quantify network similarity and identify attentional fluctuations between different cognitive states.

q-bio.NC↗

arXiv2025

Inverse Design of Chiral Structures for Giant Helical Dichroism

Investigating chiral light-matter interactions is essential for advancing applications in sensing, imaging, and pharmaceutical development. However, the chiroptical response in natural chiral molecules and subwavelength chiral structures is inherently weak, with the characterization tool limited to optical methods that utilize the light with spin angular momentum (SAM). To overcome this, orbital angular momentum (OAM) beams, characterized by helical wavefronts, have emerged as a compelling research focus. Helical dichroism (HD) describes the differential absorbance of OAM beams with opposite signs of topological charges. By using inverse design with adjoint methods for topology optimization, we design the chiral structure optimized to increase HD response under OAM beam incidence, demonstrating a giant HD response of ~107% with topological charges $|\pm\ell|$ = 3 at the wavelength of 800 nm. This study reveals distinct helicity-dependent interactions between the structure and OAM beams, highlighting the potential for custom-tuned chiroptical responses.

physics.optics↗

arXiv2024

Facial Wrinkle Segmentation for Cosmetic Dermatology: Pretraining with Texture Map-Based Weak Supervision

Facial wrinkle detection plays a crucial role in cosmetic dermatology. Precise manual segmentation of facial wrinkles is challenging and time-consuming, with inherent subjectivity leading to inconsistent results among graders. To address this issue, we propose two solutions. First, we build and release the first public facial wrinkle dataset, 'FFHQ-Wrinkle', an extension of the NVIDIA FFHQ dataset. It includes 1,000 images with human labels and 50,000 images with automatically generated weak labels. This dataset could serve as a foundation for the research community to develop advanced wrinkle detection algorithms. Second, we introduce a simple training strategy utilizing texture maps, applicable to various segmentation models, to detect wrinkles across the face. Our two-stage training strategy first pretrain models on a large dataset with weak labels (N=50k), or masked texture maps generated through computer vision techniques, without human intervention. We then finetune the models using human-labeled data (N=1k), which consists of manually labeled wrinkle masks. The network takes as input a combination of RGB and masked texture map of the image, comprising four channels, in finetuning. We effectively combine labels from multiple annotators to minimize subjectivity in manual labeling. Our strategies demonstrate improved segmentation performance in facial wrinkle segmentation both quantitatively and visually compared to existing pretraining methods. The dataset is available at https://github.com/labhai/ffhq-wrinkle-dataset.

cs.CV↗

arXiv2024

Spin-Orbit Coupling for Optical Vortex Generation in van der Waals Materials

An optical vortex beam has attracted significant attention across diverse applications, including optical manipulation, phase-contrast microscopy, optical communication, and quantum photonics. To utilize vortex generators for integrated photonics, researchers have developed ultra-compact vortex generators using fork gratings, metasurfaces, and integrated microcombs. However, those devices depend on costly, time-consuming nanofabrication and are constrained by the low signal-to-noise ratio due to the fabrication error. As an alternative maneuver, spin-orbit coupling has emerged as a method to obtain the vortex beam by converting spin angular momentum (SAM) without nanostructures. Here, we demonstrate the creation of an optical vortex beam using van der Waals (vdW) materials. The significantly high birefringence of vdW materials allows generations of optical vortex beams with high efficiency in a sub-wavelength thickness. In this work, we utilize an 8-um-thick hexagonal boron nitride (hBN) crystal for the creation of optical vortices carrying topological charges of +2 and -2. We also present the generation of an optical vortex beam in a 320-nm-thick MoS2 crystal with a conversion efficiency of 0.09. This study paves the way for fabrication-less and ultra-compact optical vortex generators, which can be applied for integrated photonics and large-scale vortex generator arrays.

physics.optics↗

arXiv2024

Deep-learning-driven end-to-end metalens imaging

Recent advances in metasurface lenses (metalenses) have shown great potential for opening a new era in compact imaging, photography, light detection and ranging (LiDAR), and virtual reality/augmented reality (VR/AR) applications. However, the fundamental trade-off between broadband focusing efficiency and operating bandwidth limits the performance of broadband metalenses, resulting in chromatic aberration, angular aberration, and a relatively low efficiency. In this study, a deep-learning-based image restoration framework is proposed to overcome these limitations and realize end-to-end metalens imaging, thereby achieving aberration-free full-color imaging for mass-produced metalenses with 10-mm diameter. Neural-network-assisted metalens imaging achieved a high resolution comparable to that of the ground truth image.

physics.optics↗

arXiv2024

Inverse Designed WS2 Planar Chiral Metasurface with Geometric Phase

Increasing attention is being paid to chiral metasurfaces due to their ability to selectively manipulate right-hand circularly polarized light or left-hand circularly polarized light. The thin nature of metasurfaces, however, poses a challenge in creating a device with effective phase modulation. Plasmonic chiral metasurfaces have attempted to address this issue by increasing light-matter interaction, but they suffer from metallic loss. Dielectric metasurfaces made from high index materials enable phase modulation while being thin. Very few materials, however, have high refractive index and low loss at visible wavelengths. Recently, some 2D materials have been shown to exhibit high refractive index and low loss in the visible wavelengths, positioning them as promising platform for meta-optics. This study introduces and details a planar chiral metasurface with geometric phase composed of WS2 meta-units. By employing adjoint optimization techniques, we achieved broadband circular dichroism.

physics.optics↗

arXiv2023

A3SA: Advanced Data Augmentation via Adjoint Sensitivity Analysis

Innovative machine learning techniques have facilitated the inverse design of photonic structures for numerous practical applications. Nevertheless, within these approaches, the quantity of data and the initial data distribution are paramount for the discovery of highly efficient photonic devices. These devices often require simulated data ranging from thousands to several hundred thousand data points. This issue has consistently posed a major hurdle in machine learning-based photonic design problems. Therefore, we propose a novel data augmentation algorithm grounded in the adjoint method, capable of generating more than 300 times the amount of original data while enhancing device efficiency. The adjoint method forecasts changes in the figure of merit (FoM) resulting from structural perturbations, requiring only two full-wave Maxwell simulations for this prediction. By leveraging the adjoint gradient values, we can augment and label several thousand new data points without any additional computations. Furthermore, the augmented data generated by the proposed algorithm displays significantly improved FoMs owing to the precise FoM change predictions enabled by the adjoint gradients. We apply this algorithm to a multi-layered metalens design problem and demonstrate that it consequently exhibits a 343-fold increase in data generation efficiency. After incorporating the proposed algorithm into a generative adversarial network (GAN), the optimized metalens exhibits a maximum focusing efficiency of 92.93%, comparable to the theoretical upper bound (93.80%).

physics.optics↗

arXiv2023

Inverse design and optical vortex manipulation for thin film absorption enhancement

Optical vortices (OVs) have rapidly varying spatial phase and optical energy that circulates around points or lines of zero optical intensity. Manipulation of OV offers innovative approaches for various fields, such as optical sensing, communication, and imaging. In this work, we demonstrate the correlation between OVs and absorption enhancement in two types of structures. First, we introduce a simple planar one-dimensional (1D) structure that manipulates OVs using two coherent light sources. The structure shows a maximum of 6.05-fold absorption gap depending on the presence of OVs. Even a slight difference in the incidence angle can influence the generation/annihilation of OVs, which implies the high sensitivity of angular light detection. Second, we apply inverse design to optimize two-dimensional (2D) perfect ultrathin absorbers. The optimized free-form structure achieves 99.90% absorptance, and the fabricable grating structure achieves 97.85% at 775 nm wavelength. To evaluate OV fields and their contribution to achieving absorption enhancement, we introduce a new parameter, OV circularity. The optimized structures generate numerous OVs with a maximum circularity of 95.37% (free-form) and 96.14% (grating), superior to our 1D structure. Our study reveals the role of high-circularity localized OVs in optimizing nano-structured absorbers and devices for optical sensing, optical communication, and many other applications.

physics.optics↗

arXiv2023

Design parameters of free-form color routers for subwavelength pixelated CMOS image sensors

Metasurface-based color routers are emerging as next-generation optical components for image sensors, replacing classical color filters and microlens arrays. In this work, we report how the design parameters such as the device dimensions and refractive indices of the dielectrics affect the optical efficiency of the color routers. Also, we report how the design grid resolution parameters affect the optical efficiency and discover that the fabrication of a color router is possible even in legacy fabrication facilities with low structure resolutions.

physics.optics↗

arXiv2022

Inverse Design of High-NA Metalens for Maskless Lithography

We demonstrate an axisymmetric inverse-designed metalens to improve the performance of zone-plate-array lithography (ZPAL), one of the maskless lithography approaches, that offer a new paradigm for nanoscale research and industry. First, we derive a computational upper bound for a unit-cell-based axisymmetric metalens. Then, we demonstrate a fabrication-compatible inverse-designed metalens with 85.50\% transmission normalized focusing efficiency at 0.6 numerical aperture at 405nm wavelength; a higher efficiency than a theoretical gradient index lens design (79.98\%). We also demonstrate experimental validation for our axisymmetric inverse-designed metalens via electron beam lithography. Metalens-based maskless lithography may open a new way of achieving low-cost, large-area nanofabrication.

physics.optics↗

arXiv2019

Maximal Free-Space Concentration of Electromagnetic Waves

We derive upper bounds to free-space concentration of electromagnetic waves, mapping out the limits to maximum intensity for any spot size and optical beam-shaping device. For sub-diffraction-limited optical beams, our bounds suggest the possibility for orders-of-magnitude intensity enhancements compared to existing demonstrations, and we use inverse design to discover metasurfaces operating near these new limits. We also demonstrate that our bounds may surprisingly describe maximum concentration defined by a wide variety of metrics. Our bounds require no assumptions about symmetry, scalar waves, or weak scattering, instead relying primarily on the transformation of a quadratic program via orthogonal-projection methods. The bounds and inverse-designed structures presented here can be useful for applications from imaging to 3D printing.

physics.optics↗

arXiv2019

Tunable metasurface inverse design for 80% switching efficiencies and 144$^\circ$ angular steering

Tunable metasurfaces have demonstrated the potential for dramatically enhanced functionality for applications including sensing, ranging and imaging. Liquid crystals (LCs) have fast switching speeds, low cost, and mature technological development, offering a versatile platform for electrical tunability. However, to date, electrically tunable metasurfaces are typically designed at a single operational state using physical intuition, without controlling alternate states and thus leading to limited switching efficiencies ($<30\%$) and small angular steering ($<$25$^\circ$). Here, we use large-scale computational 'inverse design' to discover high-performance designs through adjoint-based local-optimization design iterations within a global-optimization search. We study and explain the physics of these devices, which heavily rely on sophisticated resonator design to fully utilize the very small permittivity change incurred by switching the liquid-crystal voltage. The optimal devices show tunable steering angles ranging from 12$^\circ$ to 144$^\circ$ and switching efficiencies above 80%, exhibiting 6X angular improvements and 6X efficiency improvements compared to the current state-of-the-art.

physics.optics↗