arXiv subjects

Jinho Kim

Publications and source records attributed to Jinho Kim.

At least 19 recordsLinked to original sources

arXiv2026

Dual-Output Multi-Exposure HDR Reconstruction via SDR Fusion and Gain Map Inverse Tone Mapping

We propose DOME-HDR, a dual-output multi-exposure HDR reconstruction framework that jointly produces a perceptually balanced SDR image and a consistent HDR image via gain map inverse tone mapping. Given three bracketed LDR inputs, DOME-HDR first synthesizes a base SDR using a LoRA-adapted latent diffusion model. A dual cross-attention fusion module injects complementary structural and color cues from the under- and over-exposed images while anchoring on the mid exposure for stability. The synthesized SDR then guides HPGM, our HDR Prior-guided Gain Map network, to predict a spatially varying gain map for reliable dynamic-range expansion. We evaluate on Kalantari, Tel, and Challenge123 using both full-reference and no-reference metrics, where DOME-HDR achieves state-of-the-art HDR reconstruction quality; ablations further confirm the effectiveness of dual cross-attention and SDR-guided gain map estimation.

cs.CV↗

arXiv2026

Coarse-to-fine Framework for Generative MEF via Implicit Neural Representation

Multi-exposure fusion (MEF) expands the luminance range beyond what a single exposure can capture. Combining images taken at different exposure levels requires handling geometric differences while naturally merging their complementary brightness information. It often demands generative completion where details are missing. Diffusion-based generative methods address these challenges, however, they are computationally expensive and struggle to preserve fine structures in saturated regions. We propose LIIFusion, a coarse-to-fine framework that balances fusion quality and efficiency in generative MEF. The coarse stage performs low resolution generative fusion, enhanced by an adaptive exposure correction that recovers structure lost in saturated over-exposed areas. The fine stage adapts a local implicit image function into a multi-exposure fusion function: conditioned on the HR OE/UE sources and the coarse output, it queries arbitrary target coordinates and fuses source evidence regardless of the HR input resolution. LIIFusion achieves up to 3.5$\times$ speed-up over existing generative methods while maintaining or improving structural fidelity and perceptual quality. We believe this framework provides an effective pathway toward making generative MEF more practical in real-world applications.

cs.CV↗

arXiv2026

Effects of Differential Rotation on the Maximum Mass of Neutron Stars

The maximum mass of neutron stars provides a key constraint on the equation of state (EoS) of dense matter. Recent observations, including the ${\approx}2 M_{\odot}$ pulsar PSR~J0740+6620, have placed strong constraints on a large class of soft EoSs, while the possible existence of a compact object with a mass of $2.50$ - $2.67$ $M_{\odot}$ in GW 190814 further challenges our understanding of dense matter. Moreover, the inclusion of hyperonic degrees of freedom generally softens the EoS, making it difficult to support massive neutron stars even when the $2$ $M_{\odot}$ constraint is satisfied (a problem known as the hyperon puzzle). In this work, we investigate whether differential rotation can enhance the maximum mass of neutron stars constructed with an EoS including hyperons, thereby addressing the maximum-mass constraints imposed by current observations. We employ the Cook-Shapiro-Teukolsky (CST) approach, a numerically improved reformulation of the Komatsu-Eriguchi-Hachisu (KEH) scheme, to construct equilibrium configurations of differentially rotating neutron stars. For the nuclear matter EoS, we adopt a relativistic mean-field (RMF) model incorporating hyperonic degrees of freedom through an SU(6) symmetric coupling scheme. We find that differential rotation can substantially increase the maximum mass, yielding configurations consistent with the mass range inferred from GW 190814. However, a sufficiently soft EoS fails to satisfy the constraint from PSR~J0740+6620 (346 Hz) even with differential rotation applied. We also present a systematic analysis of the internal structure of the resulting equilibrium configurations. Furthermore, we demonstrate the existence of quasi-toroidal configurations and present equilibrium sequences incorporating the full baryon octet under extreme differential rotation.

astro-ph.HE↗

arXiv2026

Guidelines for Designing AI Technologies to Support Adult Learning

AI-powered educational technologies have demonstrated measurable benefits for learners, but their design and evaluation have largely centered on K-12 contexts. As a result, many AI-supported learning systems remain poorly aligned with the needs, constraints, and goals of adult learners. To better understand how AI systems function in adult education, this paper examines the deployment of several AI learning technologies developed within a multidisciplinary, national research institute in the United States focused on adult learning and online education. Drawing on longitudinal deployment data, we conducted a reflexive thematic analysis to identify recurring challenges and design considerations across systems. These insights were synthesized into a set of 19 design guidelines intended to inform future AI-supported adult learning technologies. We demonstrate the utility of these guidelines through a heuristic evaluation of the deployed systems. Lastly, we present a guideline exploration tool that aids in the ideation of technologies by connecting the guidelines to stakeholder statements surfaced in the analysis process.

cs.CY↗

arXiv2025

Activation Quantization of Vision Encoders Needs Prefixing Registers

Large pretrained vision encoders are central to multimodal intelligence, powering applications from on-device vision processing to vision-language models. Since these applications often demand real-time processing of massive visual data, reducing the inference cost of vision encoders is critical. Quantization offers a practical path, but it remains challenging even at 8-bit precision due to so-called outliers. In this work, we propose $\textit{RegCache}$, a training-free algorithm that mitigates outliers in large-scale pretrained vision encoders and serves as a plug-in module that can be applied on top of other quantization methods. RegCache introduces outlier-prone yet semantically meaningless prefix tokens to the vision encoder, which prevent other tokens from having outliers. Notably, we observe that outliers in vision encoders behave differently from those in language models, motivating two technical innovations: middle-layer prefixing and token deletion. Experimental results show that our method consistently improves quantized model performance across various vision encoders, particularly in extremely low-bit regimes (e.g., 4-bit).

cs.LG↗

arXiv2025

Zero-shot self-supervised learning of single breath-hold magnetic resonance cholangiopancreatography (MRCP) reconstruction

To investigate the feasibility of zero-shot self-supervised learning reconstruction for reducing breath-hold times in magnetic resonance cholangiopancreatography (MRCP). Breath-hold MRCP was acquired from 11 healthy volunteers on 3T scanners using an incoherent k-space sampling pattern, leading to 14-second acquisition time and an acceleration factor of R=25. Zero-shot reconstruction was compared with parallel imaging of respiratory-triggered MRCP (338s, R=3) and compressed sensing reconstruction. For two volunteers, breath-hold scans (40s, R=6) were additionally acquired and retrospectively undersampled to R=25 to compute peak signal-to-noise ratio (PSNR). To address long zero-shot training time, the n+m full stages of the zero-shot learning were divided into two parts to reduce backpropagation depth during training: 1) n frozen stages initialized with n-stage pretrained network and 2) m trainable stages initialized either randomly or m-stage pretrained network. Efficiency of our approach was assessed by varying initialization strategies and the number of trainable stages using the retrospectively undersampled data. Zero-shot reconstruction significantly improved visual image quality over compressed sensing, particularly in SNR and ductal delineation, and achieved image quality comparable to that of successful respiratory-triggered acquisitions with regular breathing patterns. Improved initializations enhanced PSNR and reduced reconstruction time. Adjusting frozen/trainable configurations demonstrated that PSNR decreased only slightly from 38.25 dB (0/13) to 37.67 dB (12/1), while training time decreased up to 6.7-fold. Zero-shot learning delivers high-fidelity MRCP reconstructions with reduced breath-hold times, and the proposed partially trainable approach offers a practical solution for translation into time-constrained clinical workflows.

eess.IV↗

arXiv2024

CUE-M: Contextual Understanding and Enhanced Search with Multimodal Large Language Model

The integration of Retrieval-Augmented Generation (RAG) with Multimodal Large Language Models (MLLMs) has revolutionized information retrieval and expanded the practical applications of AI. However, current systems struggle in accurately interpreting user intent, employing diverse retrieval strategies, and effectively filtering unintended or inappropriate responses, limiting their effectiveness. This paper introduces Contextual Understanding and Enhanced Search with MLLM (CUE-M), a novel multimodal search framework that addresses these challenges through a multi-stage pipeline comprising image context enrichment, intent refinement, contextual query generation, external API integration, and relevance-based filtering. CUE-M incorporates a robust filtering pipeline combining image-based, text-based, and multimodal classifiers, dynamically adapting to instance- and category-specific concern defined by organizational policies. Extensive experiments on real-word datasets and public benchmarks on knowledge-based VQA and safety demonstrated that CUE-M outperforms baselines and establishes new state-of-the-art results, advancing the capabilities of multimodal retrieval systems.

cs.CL↗

arXiv2024

Transparent Networks for Multivariate Time Series

Transparent models, which provide inherently interpretable predictions, are receiving significant attention in high-stakes domains. However, despite much real-world data being collected as time series, there is a lack of studies on transparent time series models. To address this gap, we propose a novel transparent neural network model for time series called Generalized Additive Time Series Model (GATSM). GATSM consists of two parts: 1) independent feature networks to learn feature representations, and 2) a transparent temporal module to learn temporal patterns across different time steps using the feature representations. This structure allows GATSM to effectively capture temporal patterns and handle varying-length time series while preserving transparency. Empirical experiments show that GATSM significantly outperforms existing generalized additive models and achieves comparable performance to black-box time series models, such as recurrent neural networks and Transformer. In addition, we demonstrate that GATSM finds interesting patterns in time series.

cs.LG↗

arXiv2024

General relativistic self-gravitating equilibrium disks around rotating neutron stars

In modeling a relativistic disk around a compact object, the self-gravity of the disk is often neglected while it needs to be incorporated for more accurate descriptions in several circumstances. Extending the Komatsu-Eriguchi-Hachisu self-consistent field method, we present numerical models of a rapidly rotating neutron star with a self-gravitating disk in stationary equilibrium. In particular, our approach allows us to obtain numerical solutions involving a massive disk with the rest mass $O(10^{-1})-O(10^0) M_\odot$ closely attached to a rotating neutron star. We also assess the impact of self-gravity on the internal structure of the disk and the neutron star. These axisymmetric, stationary solutions can be employed for simulations involving the neutron star-disk system in the context of high-energy transients and gravitational wave emissions.

astro-ph.HE↗

arXiv2024

Deep Learning-based Accelerated MR Cholangiopancreatography without Fully-sampled Data

The purpose of this study was to accelerate MR cholangiopancreatography (MRCP) acquisitions using deep learning-based (DL) reconstruction at 3T and 0.55T. A total of 35 healthy volunteers underwent conventional two-fold accelerated MRCP scans at field strengths of 3T and 0.55T. We trained DL reconstructions using two different training strategies, supervised (SV) and self-supervised (SSV), with retrospectively six-fold undersampled data obtained at 3T. We then evaluated the DL reconstructions against standard techniques, parallel imaging (PI) and compressed sensing (CS), focusing on peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) as metrics. We also tested DL reconstructions with prospectively accelerated acquisitions and evaluated their robustness when changing fields strengths from 3T to 0.55T. DL reconstructions demonstrated a reduction in average acquisition time from 599/542 to 255/180 seconds for MRCP at 3T/0.55T. In both retrospective and prospective undersampling, PSNR and SSIM of DL reconstructions were higher than those of PI and CS. At the same time, DL reconstructions preserved the image quality of undersampled data, including sharpness and the visibility of hepatobiliary ducts. In addition, both DL approaches produced high-quality reconstructions at 0.55T. In summary, DL reconstructions trained for highly accelerated MRCP enabled a reduction in acquisition time by a factor of 2.4/3.0 at 3T/0.55T while maintaining the image quality of conventional acquisitions.

eess.IV↗

arXiv2023

StableFDG: Style and Attention Based Learning for Federated Domain Generalization

Traditional federated learning (FL) algorithms operate under the assumption that the data distributions at training (source domains) and testing (target domain) are the same. The fact that domain shifts often occur in practice necessitates equipping FL methods with a domain generalization (DG) capability. However, existing DG algorithms face fundamental challenges in FL setups due to the lack of samples/domains in each client's local dataset. In this paper, we propose StableFDG, a style and attention based learning strategy for accomplishing federated domain generalization, introducing two key contributions. The first is style-based learning, which enables each client to explore novel styles beyond the original source domains in its local dataset, improving domain diversity based on the proposed style sharing, shifting, and exploration strategies. Our second contribution is an attention-based feature highlighter, which captures the similarities between the features of data samples in the same class, and emphasizes the important/common characteristics to better learn the domain-invariant characteristics of each class in data-poor FL scenarios. Experimental results show that StableFDG outperforms existing baselines on various DG benchmark datasets, demonstrating its efficacy.

cs.LG↗

arXiv2023

Nonexistence of colimits in naive discrete homotopy theory

We show that the quasicategory defined as the localization of the category of (simple) graphs at the class of A-homotopy equivalences does not admit colimits. In particular, we settle in the negative the question of whether the A-homotopy equivalences in the category of graphs are part of a model structure.

math.CO↗

arXiv2022

Three Dimensional Simulations of Advective, Sub-Keplerian Accretion Flow onto Non-rotating Black Holes

We study the time evolution of sub-Keplerian transonic accretion flow onto a non-rotating black hole using a three-dimensional, inviscid hydrodynamics simulation code. Prior two-dimensional simulations show that centrifugal barrier in the accreting matter may temporarily halt the nearly free-falling matter and produce a stable, geometrically thick disk which may contain turbulent eddies. Our goal in this work is to investigate whether the disk develops any instability because of this turbulence when we dynamically activate all three dimensions. We find that the disk remains stable and axisymmetric even close to the central black hole. However, if we explicitly apply non-axisymmetric azimuthal perturbation, the axisymmetric structure of the disk is destroyed and instability is developed.

astro-ph.HE↗

arXiv2022

Higher-order Neural Additive Models: An Interpretable Machine Learning Model with Feature Interactions

Neural Additive Models (NAMs) have recently demonstrated promising predictive performance while maintaining interpretability. However, their capacity is limited to capturing only first-order feature interactions, which restricts their effectiveness on real-world datasets. To address this limitation, we propose Higher-order Neural Additive Models (HONAMs), an interpretable machine learning model that effectively and efficiently captures feature interactions of arbitrary orders. HONAMs improve predictive accuracy without compromising interpretability, an essential requirement in high-stakes applications. This advantage of HONAM can help analyze and extract high-order interactions present in datasets. The source code for HONAM is publicly available at https://github.com/gim4855744/HONAM/.

cs.LG↗

arXiv2022

Explicit Feature Interaction-aware Graph Neural Networks

Graph neural networks (GNNs) are powerful tools for handling graph-structured data. However, their design often limits them to learning only higher-order feature interactions, leaving low-order feature interactions overlooked. To address this problem, we introduce a novel GNN method called explicit feature interaction-aware graph neural network (EFI-GNN). Unlike conventional GNNs, EFI-GNN is a multilayer linear network designed to model arbitrary-order feature interactions explicitly within graphs. To validate the efficacy of EFI-GNN, we conduct experiments using various datasets. The experimental results demonstrate that EFI-GNN has competitive performance with existing GNNs, and when a GNN is jointly trained with EFI-GNN, predictive performance sees an improvement. Furthermore, the predictions made by EFI-GNN are interpretable, owing to its linear construction. The source code of EFI-GNN is available at https://github.com/gim4855744/EFI-GNN

cs.LG↗

arXiv2020

Comparing Feedback Linearization and Adaptive Backstepping Control for Airborne Orientation of Agile Ground Robots using Wheel Reaction Torque

In this paper, two nonlinear methods for stabilizing the orientation of a Four-Wheel Independent Drive and Steering (4WIDS) robot while in the air are analyzed, implemented in simulation, and compared. AGRO (the Agile Ground Robot) is a 4WIDS inspection robot that can be deployed into unsafe environments by being thrown, and can use the reaction torque from its four wheels to command its orientation while in the air. Prior work has demonstrated on a hardware prototype that simple PD control with hand-tuned gains is sufficient, but hardly optimal, to stabilize the orientation in under 500ms. The goal of this work is to decrease the stabilization time and reject disturbances using nonlinear control methods. A model-based Feedback Linearization (FL) was added to compensate for the nonlinear Coriolis terms. However, with external disturbances, model uncertainty and sensor noise, the FL controller does not guarantee stability. As an alternative, a second controller was developed using backstepping methods with an adaptive compensator for external disturbances, model uncertainty, and sensor offset. The controller was designed using Lyapunov analysis. A simulation was written using the full nonlinear dynamics of AGRO in an isotropic steering configuration in which control authority over its pitch and roll are equalized. The PD+FL control method was compared to the backstepping control method using the same initial conditions in simulation. Both the backstepping controller and the PD+FL controller stabilized the system within 250 milliseconds. The adaptive backstepping controller was also able to achieve this performance with the adaptation law enabled and compensating for offset noisy sinusoidal disturbances.

cs.RO↗

arXiv2020

A Comprehensive Survey of Control Strategies for Autonomous Quadrotors

Over the past several decades there has been a constant increase in the use of Unmanned Aerial Systems (UAS). Hence, there has also been a growth in the number of control algorithms to service the many applications embodied by these vehicles. Initially UAS were made popular by the military for Reconnaissance, Intelligence, Surveillance, and Target Acquisition (RISTA) applications. Now-a-days UAS are used for everything from crop surveys to tourism. Nowhere is this more evident than with multi-rotor Unmanned Aerial Vehicle (UAV). This paper presents a survey of control methods for multi-rotor systems, namely quadrotors. In doing so, we hope to visualize a clear path to what additional capabilities might be needed in the future. In our examination, we review many of the notable research organizations and their efforts to expand the utility of multirotor aircraft. We also summarize the basic literature definitions and control strategies for autonomous quadrotors.

eess.SY↗

arXiv2019

Effects of Magnetic Field Loops on the Dynamics of Advective Accretion Flows and Jets around a Schwarzschild Blackhole

Magnetic fields advected along with low angular momentum accretion flows predominantly become toroidal due to the strong azimuthal velocity close to a black hole. We study self-consistently the movements of these flux tubes inside an advective disc and how they dynamically influence the flow. We find that the centrifugal barrier slows down the radial motion of the flux tubes. In this case, the large magnetic flux tubes with a significant drag force escape along the vertical axis due to buoyancy. Magnetic pressure rises close to the black hole and together with the centrifugal force, it combats gravity. The tug-of-war among these forces causes the centrifugal pressure supported shock to oscillate radially. We study the effects of successive injection of flux tubes and find how the flux tube could be trapped inside the disc in regions of highest entropy. Most interestingly, the shock wave remains at its average location and is not destroyed. We show that the toroidal field loops contribute significantly to collimate and accelerate the outflows from the centrifugal barrier and suggest this mechanism to be a way to collimate and accelerate jets.

astro-ph.HE↗