arXiv subjects

Junghwan Kim

Publications and source records attributed to Junghwan Kim.

At least 19 recordsLinked to original sources

arXiv2026

FinRED: An Expert-Guided Benchmark Generation and Evaluation Framework for Financial LLM Red-Teaming

Existing safety benchmarks target general adversarial scenarios but miss finance-specific risks. Financial LLMs face regulatory compliance violations, fraud facilitation, and systemic trust erosion that require targeted evaluation. We introduce FinRED, an expert-guided red-teaming framework for financial LLM safety evaluation developed with financial experts. FinRED uses a novel two-level taxonomy mapping global standards (e.g., FATF and EU DORA) to threats ranging from regulatory evasion to complex fraud, integrated with a scalable pipeline that converts real financial documents into context-rich red-teaming Behavioral Prompts (seeds) through an expert-defined schema. Rigorous expert validation confirms seed plausibility and realism for meaningful LLM safety evaluation. We also provide an expert-validated, finance-specific rubric that goes beyond disclaimer checks, aligns more closely with human experts than static one-size-fits-all rubrics, and reduces critical false negatives from 28 to 12. Aligned with internationally adopted risk-management and information-security standards (e.g., ISO/IEC 27001), FinRED is deployed in South Korea's Financial Security Institute (FSI) regulatory sandbox for generative AI security evaluation in real financial services. To mitigate dual-use risks, the dataset, generation pipeline, prompt template, and evaluation framework are gated for qualified researchers at https://github.com/selectstar-ai/FinRED-paper and https://huggingface.co/datasets/datumo/FinRED.

cs.CR↗

arXiv2026

Interpreting Style Representations via Style-Eliciting Prompts

Style representation learning is a powerful tool for authorship analysis and modeling writing style, yet the latent nature of learned representations makes them difficult to interpret. Recent work has attempted to explain these representations by generating natural language descriptions with large language models (LLMs) conditioned on input text. However, such descriptions are often prone to the LLM's biases and hallucinations, and they lack an explicit objective and practical utility. In this work, we propose a novel framework for interpreting style representations through style-eliciting prompts: natural language instructions designed to steer LLMs to generate text that reflects specific stylistic attributes. We curate 1,010 distinct style features spanning 26 stylistic categories and construct a dataset by prompting an LLM to generate text conditioned on these features. Using this data, we train a decoder to generate a style prompt from the style representation of the generated text. We evaluate our approach on three tasks: (1) recovering original style prompts from generated text, (2) generating text in the same style using the recovered prompts, and (3) steering LLM outputs to match the style of human-written texts. Experiments demonstrate that our method consistently outperforms strong baselines that directly prompt LLMs with target text, achieving superior performance in both style description and style imitation. These results highlight that style-eliciting prompts can provide a practical and interpretable interface to stylistic information encoded in style representations.

cs.CL↗

arXiv2026

STAR-Teaming: A Strategy-Response Multiplex Network Approach to Automated LLM Red Teaming

While Large Language Models (LLMs) are widely used, they remain susceptible to jailbreak prompts that can elicit harmful or inappropriate responses. This paper introduces STAR-Teaming, a novel black-box framework for automated red teaming that effectively generates such prompts. STAR-Teaming integrates a Multi-Agent System (MAS) with a Strategy-Response Multiplex Network and employs network-driven optimization to sample effective attack strategies. This network-based approach recasts the intractable high-dimensional embedding space into a tractable structure, yielding two key advantages: it enhances the interpretability of the LLM's strategic vulnerabilities, and it streamlines the search for effective strategies by organizing the search space into semantic communities, thereby preventing redundant exploration. Empirical results demonstrate that STAR-Teaming significantly surpasses existing methods, achieving a higher attack success rate (ASR) at a lower computational cost. Extensive experiments validate the effectiveness and explainability of the Multiplex Network. The code is available at https://github.com/selectstar-ai/STAR-Teaming-paper.

cs.CL↗

arXiv2026

Energy renormalizations of resident carriers and excitons in transition metal dichalcogenide monolayers

Energy renormalizations of resident carriers and excitons are studied theoretically, and compared with recent experiments of electrostatically-doped WSe$_2$ monolayers. The calculated energy renormalization of resident carriers, subjected to strong out-of-plane magnetic field, reveals the importance of dynamical screening in transition metal dichalcogenides. The energy renormalization of tightly bound excitons is analyzed through the exchange interaction between the electron (or hole) component of the exciton and resident carriers that share the same spin and valley quantum numbers. Our theory explains the weak energy shift of excitonic resonances despite the strong energy renormalization of resident carriers. We identify the dependence of the energy renormalization on the envelope function of a tightly-bound exciton, showing that unlike free electron-hole pairs, this energy renormalization is not the added renormalizations of a resident electron and resident hole.

cond-mat.mes-hall↗

arXiv2026

Optimal trajectory-guided stochastic co-optimization for e-fuel system design and real-time operation

E-fuels are promising long-term energy carriers supporting the net-zero transition. However, the large combinatorial design-operation spaces under renewable uncertainty make the use of mathematical programming impractical for co-optimizing e-fuel production systems. Here, we present MasCOR, a machine-learning-assisted co-optimization framework that learns from global operational trajectories. By encoding system design and renewable trends, a single MasCOR agent generalizes dynamic operation across diverse configurations and scenarios, substantially simplifying design-operation co-optimization under uncertainty. Benchmark comparisons against state-of-the-art reinforcement learning baselines demonstrate near-optimal performance, while computational costs are substantially lower than those of mathematical programming, enabling rapid parallel evaluation of designs within the co-optimization loop. This framework enables rapid screening of feasible design spaces together with corresponding operational policies. When applied to four potential European sites targeting e-methanol production, MasCOR shows that most locations benefit from reducing system load below 50 MW to achieve carbon-neutral methanol production, with production costs of 1.0-1.2 USD per kg. In contrast, Dunkirk (France), with limited renewable availability and high grid prices, favors system loads above 200 MW and expanded storage to exploit dynamic grid exchange and hydrogen sales to the market. These results underscore the value of the MasCOR framework for site-specific guidance from system design to real-time operation.

cs.LG↗

arXiv2026

CAGE: A Framework for Culturally Adaptive Red-Teaming Benchmark Generation

Existing red-teaming benchmarks, when adapted to new languages via direct translation, fail to capture socio-technical vulnerabilities rooted in local culture and law, creating a critical blind spot in LLM safety evaluation. To address this gap, we introduce CAGE (Culturally Adaptive Generation), a framework that systematically adapts the adversarial intent of proven red-teaming prompts to new cultural contexts. At the core of CAGE is the Semantic Mold, a novel approach that disentangles a prompt's adversarial structure from its cultural content. This approach enables the modeling of realistic, localized threats rather than testing for simple jailbreaks. As a representative example, we demonstrate our framework by creating KoRSET, a Korean benchmark, which proves more effective at revealing vulnerabilities than direct translation baselines. CAGE offers a scalable solution for developing meaningful, context-aware safety benchmarks across diverse cultures. Our dataset and evaluation rubrics are publicly available at https://github.com/selectstar-ai/CAGE-paper. (WARNING: This paper contains model outputs that can be offensive in nature.)

cs.CY↗

arXiv2026

Grounded but Misleading: Evaluating Semantic Alignment in AI-Generated Security Explanations

Online scams increasingly leverage fluent and context-aware social engineering strategies, creating growing demand for AI systems that explain why a message may be risky. However, explanations that cite detector-derived evidence may still semantically weaken or redirect the intended risk interpretation. We introduce VEXA: Verifying Semantic Explanation Alignment, a controlled testbed for studying the gap between lexical grounding and semantic risk alignment in AI-generated scam-risk explanations. VEXA generates ungrounded, risk-aligned, and risk-diluting explanations by independently controlling evidence grounding and semantic framing. Through LLM-as-a-judge and human evaluations, we show that explanations may continue to appear comparatively grounded even when their semantic interpretation weakens the detector's intended risk assessment. In human evaluation, risk-diluting XAI-grounded explanations retained comparatively elevated Perceived Evidence Grounding scores (3.66) despite lower Helpfulness (3.00) and Reasoning Support (3.14) scores. These findings provide controlled evidence of grounding illusion effects in AI-generated security explanations and suggest that trustworthy explanation evaluation must verify not only whether evidence is cited, but also how that evidence is interpreted.

cs.CR↗

arXiv2025

High luminescence efficiency of multi-valley excitonic complexes in heavily doped WSe2 monolayer

Monolayers of group-VI transition-metal dichalcogenides (TMDs) are two-dimensional semiconductors that exhibit exceptionally strong light-matter coupling yet typically suffer from low emission quantum yields. In this letter, we investigate the heavily n-doped regime of a WSe$_2$ monolayer and show that multi-particle excitonic complexes produce photoluminescence signals up to two orders of magnitude stronger than in the neutral state. Time-resolved photoluminescence and differential reflectivity measurements reveal that the quantum yield rises with carrier density and exceeds 50% for electron concentrations above 10$^{13}$ cm$^{-2}$. These findings establish TMD monolayers as a platform for exploring excitonic complexes in high-density electron gases and point toward new opportunities for efficient, atomically thin light emitters.

cond-mat.mes-hall↗

arXiv2025

Excitons in moir\'{e} superlattices with disordered electrons

Moir\'{e} superlattices in transition metal dichalcogenides (TMDs) heterobilayers exhibit various correlated insulating states driven by long-range Coulomb interactions, and these states crucially alter exciton resonances, particularly at fractional fillings. We revisit a theoretical framework to investigate the doping dependence of exciton spectra by extending hydrogenic exciton wavefunctions, systematically analyzing how the 1$s$, 2$s$, and 3$s$ Rydberg states respond to moir\'e-induced mixing of $s$- and $p$-type orbitals. Notably, while the 1$s$ state remains relatively robust against doping, higher Rydberg excitons show strong redshifts and oscillator-strength quenching near specific fractional fillings. We incorporate both defect-induced quasi-ordering and thermal fluctuations to capture realistic device conditions, employing a large supercell approach. By selectively randomizing a subset of electrons or utilizing classical Monte Carlo simulations, we present direct calculations of exciton spectra under varying defect densities and temperatures. Our results emphasize how even moderate disorder or finite temperature can partially or completely suppress characteristic moir\'{e} exciton physics. Especially, we show how the 2$s$ exciton states respond to the phase transition in correlated electron states. This comprehensive picture not only clarifies recent experimental observations but also provides a framework to guide the design of moir\'{e}-based optoelectronic devices.

cond-mat.mtrl-sci↗

arXiv2025

Leveraging Multilingual Training for Authorship Representation: Enhancing Generalization across Languages and Domains

Authorship representation (AR) learning, which models an author's unique writing style, has demonstrated strong performance in authorship attribution tasks. However, prior research has primarily focused on monolingual settings-mostly in English-leaving the potential benefits of multilingual AR models underexplored. We introduce a novel method for multilingual AR learning that incorporates two key innovations: probabilistic content masking, which encourages the model to focus on stylistically indicative words rather than content-specific words, and language-aware batching, which improves contrastive learning by reducing cross-lingual interference. Our model is trained on over 4.5 million authors across 36 languages and 13 domains. It consistently outperforms monolingual baselines in 21 out of 22 non-English languages, achieving an average Recall@8 improvement of 4.85%, with a maximum gain of 15.91% in a single language. Furthermore, it exhibits stronger cross-lingual and cross-domain generalization compared to a monolingual model trained solely on English. Our analysis confirms the effectiveness of both proposed techniques, highlighting their critical roles in the model's improved performance.

cs.CL↗

arXiv2025

MAS-LitEval : Multi-Agent System for Literary Translation Quality Assessment

Literary translation requires preserving cultural nuances and stylistic elements, which traditional metrics like BLEU and METEOR fail to assess due to their focus on lexical overlap. This oversight neglects the narrative consistency and stylistic fidelity that are crucial for literary works. To address this, we propose MAS-LitEval, a multi-agent system using Large Language Models (LLMs) to evaluate translations based on terminology, narrative, and style. We tested MAS-LitEval on translations of The Little Prince and A Connecticut Yankee in King Arthur's Court, generated by various LLMs, and compared it to traditional metrics. \textbf{MAS-LitEval} outperformed these metrics, with top models scoring up to 0.890 in capturing literary nuances. This work introduces a scalable, nuanced framework for Translation Quality Assessment (TQA), offering a practical tool for translators and researchers.

cs.CL↗

arXiv2025

Ten-valley excitonic complexes in charge-tunable monolayer WSe$_2$

Excitons dominate the optical response of two-dimensional (2D) semiconductors. Strong interactions produce peculiar excitonic complexes, which provide a testing ground for exciton and quantum many-body theories. Here, we report a hitherto unobserved many-body exciton that emerges upon filling both the K and Q valleys of WSe$_2$. We optically probe the exciton landscape using charge-tunable devices with unusually thin dielectrics that facilitate doping up to several $10^{13}$ cm$^{-2}$. We observe the emergence of the thermodynamically stable complex when 10 valleys are electrostatically filled. We gain insight into its physics using magneto-optical measurements. Our results are well-described by a model where the number of distinguishable Fermi seas interacting with the photoexcited electron-hole pair defines the complex's behavior. In addition to expanding the repertoire of excitons in 2D semiconductors, this complex could probe the limit of exciton models and answer open questions about screened Coulomb interactions in 2D semiconductors.

cond-mat.mes-hall↗

arXiv2025

Excitons in fractionally-filled moir\'{e} superlattices

Long-range Coulomb forces give rise to correlated insulating states when charge particles populate a moir\'{e} superlattice at certain fractional filling factors. Such behavior is characterized by a broken translation symmetry wherein particles spontaneously form a Wigner crystal. Focusing on the experimental findings of Xu et al. [Nature \textbf{587}, 214 (2020)], we present a theory that captures the correlated insulating state of a fractionally-filled moir\'{e} superlattice through the energy shift and change in oscillator strength of the exciton absorption resonance. The theory shows that the experimental findings can only be supported if the electrons reside in a charge-ordered state (i.e., electrons are not randomly distributed among the sites of the moir\'{e} superlattice). Furthermore, we explain why the energy shifts of exciton resonances are qualitatively different in cases that the superlattice is nearly empty compared with a superlattice whose sites are doubly occupied.

cond-mat.mtrl-sci↗

arXiv2025

Assessing Computer Science Student Attitudes Towards AI Ethics and Policy

As artificial intelligence (AI) grows in popularity and importance-both as a domain within broader computing research and in society at large-increasing focus will need to be paid to the ethical governance of this emerging technology. The attitudes and competencies with respect to AI ethics and policy among post-secondary students studying computer science (CS) are of particular interest, as many of these students will go on to play key roles in the development and deployment of future AI innovations. Despite this population of computer scientists being at the forefront of learning about and using AI tools, their attitudes towards AI remain understudied in the literature. In an effort to begin to close this gap, in fall 2024 we fielded a survey ($n=117$) to undergraduate and graduate students enrolled in CS courses at a large public university in the United States to assess their attitudes towards the nascent fields of AI ethics and policy. Additionally, we conducted one-on-one follow-up interviews with 13 students to elicit more in-depth responses on topics such as the use of AI tools in the classroom, ethical impacts of AI, and government regulation of AI. In this paper, we describe the findings of both the survey and interviews, drawing parallels and contrasts to broader public opinion polling in the United States. We conclude by evaluating the implications of CS student attitudes on the future of AI education and governance.

cs.CY↗

arXiv2025

Optimizing 4D Gaussians for Dynamic Scene Video from Single Landscape Images

To achieve realistic immersion in landscape images, fluids such as water and clouds need to move within the image while revealing new scenes from various camera perspectives. Recently, a field called dynamic scene video has emerged, which combines single image animation with 3D photography. These methods use pseudo 3D space, implicitly represented with Layered Depth Images (LDIs). LDIs separate a single image into depth-based layers, which enables elements like water and clouds to move within the image while revealing new scenes from different camera perspectives. However, as landscapes typically consist of continuous elements, including fluids, the representation of a 3D space separates a landscape image into discrete layers, and it can lead to diminished depth perception and potential distortions depending on camera movement. Furthermore, due to its implicit modeling of 3D space, the output may be limited to videos in the 2D domain, potentially reducing their versatility. In this paper, we propose representing a complete 3D space for dynamic scene video by modeling explicit representations, specifically 4D Gaussians, from a single image. The framework is focused on optimizing 3D Gaussians by generating multi-view images from a single image and creating 3D motion to optimize 4D Gaussians. The most important part of proposed framework is consistent 3D motion estimation, which estimates common motion among multi-view images to bring the motion in 3D space closer to actual motions. As far as we know, this is the first attempt that considers animation while representing a complete 3D space from a single landscape image. Our model demonstrates the ability to provide realistic immersion in various landscape images through diverse experiments and metrics. Extensive experimental results are https://cvsp-lab.github.io/ICLR2025_3D-MOM/.

cs.CV↗

arXiv2025

LLMDR: LLM-Driven Deadlock Detection and Resolution in Multi-Agent Pathfinding

Multi-Agent Pathfinding (MAPF) is a core challenge in multi-agent systems. Existing learning-based MAPF methods often struggle with scalability, particularly when addressing complex scenarios that are prone to deadlocks. To address these challenges, we introduce LLMDR (LLM-Driven Deadlock Detection and Resolution), an approach designed to resolve deadlocks and improve the performance of learnt MAPF models. LLMDR integrates the inference capabilities of large language models (LLMs) with learnt MAPF models and prioritized planning, enabling it to detect deadlocks and provide customized resolution strategies. We evaluate LLMDR on standard MAPF benchmark maps with varying agent numbers, measuring its performance when combined with several base models. The results demonstrate that LLMDR improves the performance of learnt MAPF models, particularly in deadlock-prone scenarios, with notable improvements in success rates. These findings show the potential of integrating LLMs to improve the scalability of learning-based MAPF methods. The source code for LLMDR is available at: https://github.com/ssbacc/llmdr-dhc

cs.MA↗

arXiv2025

State-of-Health Prediction for EV Lithium-Ion Batteries via DLinear and Robust Explainable Feature Selection

Accurate prediction of the state-of-health (SOH) of lithium-ion batteries is essential for ensuring the safety, reliability, and efficient operation of electric vehicles (EVs). Battery packs in EVs experience nonuniform degradation due to cell-to-cell variability (CtCV), posing a major challenge for real-time battery management. In this work, we propose an explainable, data-driven SOH prediction framework tailored for EV battery management systems (BMS). The approach combines robust feature engineering with a DLinear. Using NASA's battery aging dataset, we extract twenty meaningful features from voltage, current, temperature, and time profiles, and select key features using Pearson correlation and Shapley additive explanations (SHAP). The SHAP-based selection yields consistent feature importance across multiple cells, effectively capturing CtCV. The DLinear algorithm outperforms long short-term memory (LSTM) and Transformer architectures in prediction accuracy, while requiring fewer training cycles and lower computational cost. This work offers a scalable and interpretable framework for SOH forecasting, enabling practical implementation in EV BMS and promoting safer, more efficient electric mobility.

eess.SY↗

arXiv2024

Real or Robotic? Assessing Whether LLMs Accurately Simulate Qualities of Human Responses in Dialogue

Studying and building datasets for dialogue tasks is both expensive and time-consuming due to the need to recruit, train, and collect data from study participants. In response, much recent work has sought to use large language models (LLMs) to simulate both human-human and human-LLM interactions, as they have been shown to generate convincingly human-like text in many settings. However, to what extent do LLM-based simulations \textit{actually} reflect human dialogues? In this work, we answer this question by generating a large-scale dataset of 100,000 paired LLM-LLM and human-LLM dialogues from the WildChat dataset and quantifying how well the LLM simulations align with their human counterparts. Overall, we find relatively low alignment between simulations and human interactions, demonstrating a systematic divergence along the multiple textual properties, including style and content. Further, in comparisons of English, Chinese, and Russian dialogues, we find that models perform similarly. Our results suggest that LLMs generally perform better when the human themself writes in a way that is more similar to the LLM's own style.

cs.CL↗