arXiv subjects

Minkyung Kim

Publications and source records attributed to Minkyung Kim.

At least 19 recordsLinked to original sources

arXiv2026

3rd Place Solution to Human Motion Challenges in Real-World and Clinical Settings (MoCha) @ECCV2026: Language-Aligned Motion Representations for Domain-Generalizable UPDRS-Gait Severity Estimation

In this work, we introduce language-aligned motion representations for domain-generalizable UPDRS-Gait severity estimation, aiming to learn semantically structured motion features that generalize across heterogeneous clinical domains. We first learn motion representations using a Bi-GRU backbone that captures the temporal dynamics of SMPL sequences. Prior to model training, motion captions are generated offline using Qwen2.5-7B-Instruct. The backbone is then trained with both classification and text-alignment objectives to learn discriminative and semantically structured motion representations while accounting for the class imbalance present in the training data. We subsequently adapt the learned backbone independently to each source domain so that the model can capture domain-specific motion characteristics. The resulting source-specific models are then merged at the parameter level to consolidate complementary knowledge across source domains into a single domain-generalized model. To further mitigate class imbalance, we perform GPT-5.5-based pseudo labeling, and our final merged models for each site do not use any class-prior correction during inference. The resulting model is evaluated under the unseen-site setting of the MoCha Challenge, using Macro F1 as the primary evaluation metric. Our method achieves a macro-F1 of 0.57 on the hidden test set with only 637K active parameters at inference, ranking 3rd among 58 leaderboard entries in the MoCha 2026 Challenge. The challenge attracted 1,669 submissions from 112 participants and offered monetary prizes sponsored by Machine Medicine Technologies.

cs.CV↗

arXiv2026

MUSE: Multimodal Uncertainty Quantification of State Estimation

Accurate visual state estimation has been a central topic in robotics with a wide range of applications in robot navigation, autonomous driving, and autonomous flight. Recent advances in robot perception have led to significant improvements in the accuracy and robustness of state estimation, yet a fundamental challenge remains in how to quantify and calibrate its precision, i.e., how confident we are in an estimate and whether failures can be detected. This issue is particularly pronounced in visual-inertial odometry (VIO), where the heteroscedastic and multimodal nature of the problem makes uncertainty quantification especially difficult. This paper introduces MUSE (Multimodal Uncertainty Quantification of State Estimation), a novel real-time learning-based framework that leverages the strong and efficient sequential modeling capacity of Mamba to estimate localization uncertainty from multiple asynchronous sensor streams. Experiments on both public and in-house datasets demonstrate that MUSE achieves superior reliability and robustness compared to existing uncertainty quantification methods, and ablation studies justify the benefits of its key design choices.

cs.RO↗

arXiv2025

Aligning Large Language Models with Procedural Rules: An Autoregressive State-Tracking Prompting for In-Game Trading

Large Language Models (LLMs) enable dynamic game interactions but fail to follow essential procedural flows in rule-governed trading systems, eroding player trust. This work resolves the core tension between the creative flexibility of LLMs and the procedural demands of in-game trading (browse-offer-review-confirm). To this end, Autoregressive State-Tracking Prompting (ASTP) is introduced, a methodology centered on a strategically orchestrated prompt that compels an LLM to make its state-tracking process explicit and verifiable. Instead of relying on implicit contextual understanding, ASTP tasks the LLM with identifying and reporting a predefined state label from the previous turn. To ensure transactional integrity, this is complemented by a state-specific placeholder post-processing method for accurate price calculations. Evaluation across 300 trading dialogues demonstrates >99% state compliance and 99.3% calculation precision. Notably, ASTP with placeholder post-processing on smaller models (Gemini-2.5-Flash) matches larger models' (Gemini-2.5-Pro) performance while reducing response time from 21.2s to 2.4s, establishing a practical foundation that satisfies both real-time requirements and resource constraints of commercial games.

cs.AI↗

arXiv2025

State-Inference-Based Prompting for Natural Language Trading with Game NPCs

Large Language Models enable dynamic game interactions but struggle with rule-governed trading systems. Current implementations suffer from rule violations, such as item hallucinations and calculation errors, that erode player trust. Here, State-Inference-Based Prompting (SIBP) enables reliable trading through autonomous dialogue state inference and context-specific rule adherence. The approach decomposes trading into six states within a unified prompt framework, implementing context-aware item referencing and placeholder-based price calculations. Evaluation across 100 trading dialogues demonstrates >97% state compliance, >95% referencing accuracy, and 99.7% calculation precision. SIBP maintains computational efficiency while outperforming baseline approaches, establishing a practical foundation for trustworthy NPC interactions in commercial games.

cs.AI↗

arXiv2025

RAW Image Reconstruction from RGB on Smartphones. NTIRE 2025 Challenge Report

Numerous low-level vision tasks operate in the RAW domain due to its linear properties, bit depth, and sensor designs. Despite this, RAW image datasets are scarce and more expensive to collect than the already large and public sRGB datasets. For this reason, many approaches try to generate realistic RAW images using sensor information and sRGB images. This paper covers the second challenge on RAW Reconstruction from sRGB (Reverse ISP). We aim to recover RAW sensor images from smartphones given the corresponding sRGB images without metadata and, by doing this, ``reverse" the ISP transformation. Over 150 participants joined this NTIRE 2025 challenge and submitted efficient models. The proposed methods and benchmark establish the state-of-the-art for generating realistic RAW data.

eess.IV↗

arXiv2025

"We're losing our neighborhoods. We're losing our community": A comparative analysis of community discourse in online and offline public spheres

Recovering from crises, such as hurricanes or wildfires, is a complex process that can take weeks, months, or even decades to overcome. Crises have both acute (immediate) and chronic (long-term) effects on communities. Crisis informatics research often focuses on the immediate response phase of disasters, thereby overlooking the long-term recovery phase, which is critical for understanding the information needs of users undergoing challenges like climate gentrification and housing inequity. We fill this gap by investigating community discourse over eight months following Hurricane Ida in an online neighborhood Facebook group and Town Hall Meetings of a borough in the New York Metropolitan region. Using a mixed methods approach, we examined the use of social media to manage long-term disaster recovery. The findings revealed a significant overlap in topics, underscoring the interconnected nature of online and offline community discourse, and illuminated themes related to the long-term consequences of disasters. We conclude with recommendations aimed at helping designers and government leaders enhance participation across community forums and support recovery in the aftermath of disasters.

cs.SI↗

arXiv2024

DiffTune: Auto-Tuning through Auto-Differentiation

The performance of robots in high-level tasks depends on the quality of their lower-level controller, which requires fine-tuning. However, the intrinsically nonlinear dynamics and controllers make tuning a challenging task when it is done by hand. In this paper, we present DiffTune, a novel, gradient-based automatic tuning framework. We formulate the controller tuning as a parameter optimization problem. Our method unrolls the dynamical system and controller as a computational graph and updates the controller parameters through gradient-based optimization. The gradient is obtained using sensitivity propagation, which is the only method for gradient computation when tuning for a physical system instead of its simulated counterpart. Furthermore, we use $\mathcal{L}_1$ adaptive control to compensate for the uncertainties (that unavoidably exist in a physical system) such that the gradient is not biased by the unmodelled uncertainties. We validate the DiffTune on a Dubin's car and a quadrotor in challenging simulation environments. In comparison with state-of-the-art auto-tuning methods, DiffTune achieves the best performance in a more efficient manner owing to its effective usage of the first-order information of the system. Experiments on tuning a nonlinear controller for quadrotor show promising results, where DiffTune achieves 3.5x tracking error reduction on an aggressive trajectory in only 10 trials over a 12-dimensional controller parameter space.

cs.RO↗

arXiv2023

Active anomaly detection based on deep one-class classification

Active learning has been utilized as an efficient tool in building anomaly detection models by leveraging expert feedback. In an active learning framework, a model queries samples to be labeled by experts and re-trains the model with the labeled data samples. It unburdens in obtaining annotated datasets while improving anomaly detection performance. However, most of the existing studies focus on helping experts identify as many abnormal data samples as possible, which is a sub-optimal approach for one-class classification-based deep anomaly detection. In this paper, we tackle two essential problems of active learning for Deep SVDD: query strategy and semi-supervised learning method. First, rather than solely identifying anomalies, our query strategy selects uncertain samples according to an adaptive boundary. Second, we apply noise contrastive estimation in training a one-class classification model to incorporate both labeled normal and abnormal data effectively. We analyze that the proposed query strategy and semi-supervised loss individually improve an active learning process of anomaly detection and further improve when combined together on seven anomaly detection datasets.

cs.LG↗

arXiv2023

An Iterative Method for Unsupervised Robust Anomaly Detection Under Data Contamination

Most deep anomaly detection models are based on learning normality from datasets due to the difficulty of defining abnormality by its diverse and inconsistent nature. Therefore, it has been a common practice to learn normality under the assumption that anomalous data are absent in a training dataset, which we call normality assumption. However, in practice, the normality assumption is often violated due to the nature of real data distributions that includes anomalous tails, i.e., a contaminated dataset. Thereby, the gap between the assumption and actual training data affects detrimentally in learning of an anomaly detection model. In this work, we propose a learning framework to reduce this gap and achieve better normality representation. Our key idea is to identify sample-wise normality and utilize it as an importance weight, which is updated iteratively during the training. Our framework is designed to be model-agnostic and hyperparameter insensitive so that it applies to a wide range of existing methods without careful parameter tuning. We apply our framework to three different representative approaches of deep anomaly detection that are classified into one-class classification-, probabilistic model-, and reconstruction-based approaches. In addition, we address the importance of a termination condition for iterative methods and propose a termination criterion inspired by the anomaly detection objective. We validate that our framework improves the robustness of the anomaly detection models under different levels of contamination ratios on five anomaly detection benchmark datasets and two image datasets. On various contaminated datasets, our framework improves the performance of three representative anomaly detection methods, measured by area under the ROC curve.

cs.LG↗

arXiv2023

DiffTune$^+$: Hyperparameter-Free Auto-Tuning using Auto-Differentiation

Controller tuning is a vital step to ensure the controller delivers its designed performance. DiffTune has been proposed as an automatic tuning method that unrolls the dynamical system and controller into a computational graph and uses auto-differentiation to obtain the gradient for the controller's parameter update. However, DiffTune uses the vanilla gradient descent to iteratively update the parameter, in which the performance largely depends on the choice of the learning rate (as a hyperparameter). In this paper, we propose to use hyperparameter-free methods to update the controller parameters. We find the optimal parameter update by maximizing the loss reduction, where a predicted loss based on the approximated state and control is used for the maximization. Two methods are proposed to optimally update the parameters and are compared with related variants in simulations on a Dubin's car and a quadrotor. Simulation experiments show that the proposed first-order method outperforms the hyperparameter-based methods and is more robust than the second-order hyperparameter-free methods.

cs.RO↗

arXiv2023

Unsupervised Deep One-Class Classification with Adaptive Threshold based on Training Dynamics

One-class classification has been a prevailing method in building deep anomaly detection models under the assumption that a dataset consisting of normal samples is available. In practice, however, abnormal samples are often mixed in a training dataset, and they detrimentally affect the training of deep models, which limits their applicability. For robust normality learning of deep practical models, we propose an unsupervised deep one-class classification that learns normality from pseudo-labeled normal samples, i.e., outlier detection in single cluster scenarios. To this end, we propose a pseudo-labeling method by an adaptive threshold selected by ranking-based training dynamics. The experiments on 10 anomaly detection benchmarks show that our method effectively improves performance on anomaly detection by sizable margins.

cs.LG↗

arXiv2021

Normality-Calibrated Autoencoder for Unsupervised Anomaly Detection on Data Contamination

In this paper, we propose Normality-Calibrated Autoencoder (NCAE), which can boost anomaly detection performance on the contaminated datasets without any prior information or explicit abnormal samples in the training phase. The NCAE adversarially generates high confident normal samples from a latent space having low entropy and leverages them to predict abnormal samples in a training dataset. NCAE is trained to minimise reconstruction errors in uncontaminated samples and maximise reconstruction errors in contaminated samples. The experimental results demonstrate that our method outperforms shallow, hybrid, and deep methods for unsupervised anomaly detection and achieves comparable performance compared with semi-supervised methods using labelled anomaly samples in the training phase. The source code is publicly available on `https://github.com/andreYoo/NCAE_UAD.git'.

cs.LG↗

arXiv2021

Camera-Tracklet-Aware Contrastive Learning for Unsupervised Vehicle Re-Identification

Recently, vehicle re-identification methods based on deep learning constitute remarkable achievement. However, this achievement requires large-scale and well-annotated datasets. In constructing the dataset, assigning globally available identities (Ids) to vehicles captured from a great number of cameras is labour-intensive, because it needs to consider their subtle appearance differences or viewpoint variations. In this paper, we propose camera-tracklet-aware contrastive learning (CTACL) using the multi-camera tracklet information without vehicle identity labels. The proposed CTACL divides an unlabelled domain, i.e., entire vehicle images, into multiple camera-level subdomains and conducts contrastive learning within and beyond the subdomains. The positive and negative samples for contrastive learning are defined using tracklet Ids of each camera. Additionally, the domain adaptation across camera networks is introduced to improve the generalisation performance of learnt representations and alleviate the performance degradation resulted from the domain gap between the subdomains. We demonstrate the effectiveness of our approach on video-based and image-based vehicle Re-ID datasets. Experimental results show that the proposed method outperforms the recent state-of-the-art unsupervised vehicle Re-ID methods. The source code for this paper is publicly available on `https://github.com/andreYoo/CTAM-CTACL-VVReID.git'.

cs.CV↗

arXiv2021

Incident-polarization-independent spin Hall effect of light reaching a half beam waist

The spin Hall effect of light, a spin-dependent transverse splitting of light at an optical interface, is intrinsically an incident-polarization-sensitive phenomenon. Recently, an approach to eliminate the polarization dependence by equalizing the reflection coefficients of two linear polarizations has been proposed, but is only valid when the beam waist is sufficiently larger than the wavelength. Here, we demonstrate that an interface, at which the reflection coefficients of the two linear polarizations are the same and so are their derivatives with respect to the incident angle, supports the polarization-independent spin Hall shift, even when the beam waist is comparable to the wavelength. In addition, an isotropic-anisotropic interface that exhibits the polarization-independent spin Hall shift over the entire range of incident angles is presented. Monte-Carlo simulations prove that spin Hall shifts are degenerate under any polarization and reaches a half of beam waist under unpolarized incidence. We suggest an application of the beam-waist-scale spin Hall effect of light as a tunable beam-splitting device that is responsive to the incident polarization. The spin Hall shift that is independent of the incident polarization at any incident angle will facilitate a wide range of applications including practical spin-dependent devices and active beam splitters.

physics.optics↗

arXiv2021

Measuring Human Adaptation to AI in Decision Making: Application to Evaluate Changes after AlphaGo

Across a growing number of domains, human experts are expected to learn from and adapt to AI with superior decision making abilities. But how can we quantify such human adaptation to AI? We develop a simple measure of human adaptation to AI and test its usefulness in two case studies. In Study 1, we analyze 1.3 million move decisions made by professional Go players and find that a positive form of adaptation to AI (learning) occurred after the players could observe the reasoning processes of AI, rather than mere actions of AI. These findings based on our measure highlight the importance of explainability for human learning from AI. In Study 2, we test whether our measure is sufficiently sensitive to capture a negative form of adaptation to AI (cheating aided by AI), which occurred in a match between professional Go players. We discuss our measure's applications in domains other than Go, especially in domains in which AI's decision making ability will likely surpass that of human experts.

cs.HC↗

arXiv2021

Spin Hall effect of light under arbitrarily polarized and unpolarized light

The spin Hall effect of light (SHEL), which refers to a spin-dependent and transverse splitting at refraction and reflection phenomena, inherently depends on the polarization states of the incidence. Most of the previous research have focused on a horizontally or vertically polarized incidence, in which the analytic formula of the shift is well-formulated and SHEL appears symmetrically in both shift and intensity. However, the SHEL under an arbitrarily polarized or unpolarized incidence has remained largely unexplored. Whereas the SHEL under other polarization is sensitive to incident polarization and is asymmetrical, here we demonstrate that the SHEL is independent of the incident polarization and is symmetrical in shift if Fresnel coefficients of the two linear polarization are the same. The independence of the shift with respect to the incident polarization is proved both analytically and numerically. Moreover, we prove that under an unpolarized incidence composed of a large number of completely random polarization states, the reflected beam is split in half into two circularly polarized components that undergo the same amount of splitting but in opposite directions. This result means that under unpolarized incidence, the SHEL occurs exactly the same as under horizontally or vertically polarized incidence. We believe that the incident-polarization-independent SHEL can broaden the applicability of the SHEL to cover optical systems in which polarization is ill-defined.

physics.optics↗

arXiv2019

Topological edge and corner states in a two-dimensional photonic Su-Schrieffer-Heeger lattice

Implementation of topology on photonics has opened new functionalities of photonic systems such as topologically protected boundary modes. We present polarization-dependent topological properties in 2D Su-Schrieffer-Heeger lattice by using a metallic nanoparticle array and considering the polarization degree of freedom. We demonstrate that when eigenmodes are polarized parallel to the plane of the 2D lattice, it supports isolated longitudinal edge modes and transverse modes that are hidden from the projected bulk states. Also, the in-plane polarized modes support a second-order topological phase under an open boundary condition by breaking the four-fold rotational symmetry. This work will offer polarization-based multifunctionality in compact photonic systems that have topological features.

physics.optics↗

arXiv2019

Observation of enhanced optical spin Hall effect in a vertical hyperbolic metamaterial

Hyperbolic metamaterials, horizontally stacked metal and dielectric multilayer, have recently been studied as a platform to observe optical spin Hall effect. However, the large optical spin Hall effect in the horizontal hyperbolic metamaterials accompanies extremely low transmission, which obstructs its practical applications. Reducing the sample thickness to augment the transmission causes diminishment of the shift. In this letter, we demonstrate that a vertical hyperbolic metamaterial can enhance the shift by several orders of magnitude in comparison to the shift of its horizontal counterpart. Under the same conditions of material combinations and total thickness, the shift enhancement, which is incident angle-dependent, can be higher than 800-fold when the incident angle is 5 degree, and 5000-fold when the incident angle is 1 degree. As a proof of concept, we fabricate a large-scale gold nano-grating by nanoimprint lithography and measure the helicity-dependent shift by Stokes polarimetry setup, which agrees well with the simulated result. The gigantic optical spin Hall effect in a vertical hyperbolic metamaterial will enable helicity-dependent control of optical devices including filters, sensors, switches and beam splitters.

physics.optics↗