arXiv subjects

Yohan Lee

Publications and source records attributed to Yohan Lee.

13 recordsLinked to original sources

arXiv2026

Dual-Gradient Plasmonic qBIC Metasurface for Time-Resolved In Situ Optimization of Molecular Vibrational Sensing in Water

Metaphotonic platforms based on quasi-bound states in the continuum (qBICs), harnessing strong near-field enhancement and deeply subwavelength field confinement, provide a versatile framework for surface-enhanced infrared absorption (SEIRA) spectroscopy. However, real-time molecular sensing in water using qBIC metasurfaces remains challenging. First, strong water absorption in the mid-infrared region damps qBICs and obscures weak analyte signals. Second, conventional metasurfaces rely on discrete arrays targeting individual wavelengths and coupling conditions, increasing the device footprint and fabrication effort required to match the qBIC to the molecular vibrational resonance. Here, we present a dual-gradient plasmonic qBIC metasurface with diamond-shaped resonators for in situ molecular vibrational sensing in water, where spatially encoded gradients in the scaling factor and asymmetry parameter independently control the qBIC spectral positions and radiative rates, respectively, across a 1000 um x 700 um footprint. Building on this capability, we experimentally monitor lipid vesicle dynamics in real time and resolve the carbonyl vibrational signature despite the unavoidable water absorption. We then directly identify the optimal sensing condition and track its evolution across the spatially encoded parameter space. The results establish a compact single-chip strategy that combines adsorption-kinetics monitoring, vibrational fingerprint detection, and on-chip optimization, opening opportunities for investigating biological dynamics under aqueous conditions.

physics.optics↗

arXiv2026

Overthinking Loops in Agents: A Structural Risk via MCP Tools

Tool-using LLM agents increasingly coordinate real workloads by selecting and chaining third-party tools based on text-visible metadata such as tool names, descriptions, and return messages. We show that this convenience creates a supply-chain attack surface: a malicious MCP tool server can be co-registered alongside normal tools and induce overthinking loops, where individually trivial or plausible tool calls compose into cyclic trajectories that inflate end-to-end tokens and latency without any single step looking abnormal. We formalize this as a structural overthinking attack, distinguishable from token-level verbosity, and implement 14 malicious tools across three servers that trigger repetition, forced refinement, and distraction. Across heterogeneous registries and multiple tool-capable models, the attack causes severe resource amplification (up to $142.4\times$ tokens) and can degrade task outcomes. Finally, we find that decoding-time concision controls do not reliably prevent loop induction, suggesting defenses should reason about tool-call structure rather than tokens alone.

cs.CL↗

arXiv2025

Chronic Kidney Disease Prognosis Prediction Using Transformer

Chronic Kidney Disease (CKD) affects nearly 10\% of the global population and often progresses to end-stage renal failure. Accurate prognosis prediction is vital for timely interventions and resource optimization. We present a transformer-based framework for predicting CKD progression using multi-modal electronic health records (EHR) from the Seoul National University Hospital OMOP Common Data Model. Our approach (\textbf{ProQ-BERT}) integrates demographic, clinical, and laboratory data, employing quantization-based tokenization for continuous lab values and attention mechanisms for interpretability. The model was pretrained with masked language modeling and fine-tuned for binary classification tasks predicting progression from stage 3a to stage 5 across varying follow-up and assessment periods. Evaluated on a cohort of 91,816 patients, our model consistently outperformed CEHR-BERT, achieving ROC-AUC up to 0.995 and PR-AUC up to 0.989 for short-term prediction. These results highlight the effectiveness of transformer architectures and temporal design choices in clinical prognosis modeling, offering a promising direction for personalized CKD care.

cs.AI↗

arXiv2025

Finding Diamonds in Conversation Haystacks: A Benchmark for Conversational Data Retrieval

We present the Conversational Data Retrieval (CDR) benchmark, the first comprehensive test set for evaluating systems that retrieve conversation data for product insights. With 1.6k queries across five analytical tasks and 9.1k conversations, our benchmark provides a reliable standard for measuring conversational data retrieval performance. Our evaluation of 16 popular embedding models shows that even the best models reach only around NDCG@10 of 0.51, revealing a substantial gap between document and conversational data retrieval capabilities. Our work identifies unique challenges in conversational data retrieval (implicit state recognition, turn dynamics, contextual references) while providing practical query templates and detailed error analysis across different task categories. The benchmark dataset and code are available at https://github.com/l-yohai/CDR-Benchmark.

cs.CL↗

arXiv2025

Pre-Storage Reasoning for Episodic Memory: Shifting Inference Burden to Memory for Personalized Dialogue

Effective long-term memory in conversational AI requires synthesizing information across multiple sessions. However, current systems place excessive reasoning burden on response generation, making performance significantly dependent on model sizes. We introduce PREMem (Pre-storage Reasoning for Episodic Memory), a novel approach that shifts complex reasoning processes from inference to memory construction. PREMem extracts fine-grained memory fragments categorized into factual, experiential, and subjective information; it then establishes explicit relationships between memory items across sessions, capturing evolution patterns like extensions, transformations, and implications. By performing this reasoning during pre-storage rather than when generating a response, PREMem creates enriched representations while reducing computational demands during interactions. Experiments show significant performance improvements across all model sizes, with smaller models achieving results comparable to much larger baselines while maintaining effectiveness even with constrained token budgets. Code and dataset are available at https://github.com/sangyeop-kim/PREMem.

cs.CL↗

arXiv2025

FEAT: A Preference Feedback Dataset through a Cost-Effective Auto-Generation and Labeling Framework for English AI Tutoring

In English education tutoring, teacher feedback is essential for guiding students. Recently, AI-based tutoring systems have emerged to assist teachers; however, these systems require high-quality and large-scale teacher feedback data, which is both time-consuming and costly to generate manually. In this study, we propose FEAT, a cost-effective framework for generating teacher feedback, and have constructed three complementary datasets: (1) DIRECT-Manual (DM), where both humans and large language models (LLMs) collaboratively generate high-quality teacher feedback, albeit at a higher cost; (2) DIRECT-Generated (DG), an LLM-only generated, cost-effective dataset with lower quality;, and (3) DIRECT-Augmented (DA), primarily based on DG with a small portion of DM added to enhance quality while maintaining cost-efficiency. Experimental results showed that incorporating a small portion of DM (5-10%) into DG leads to superior performance compared to using 100% DM alone.

cs.AI↗

arXiv2025

What Really Matters in Many-Shot Attacks? An Empirical Study of Long-Context Vulnerabilities in LLMs

We investigate long-context vulnerabilities in Large Language Models (LLMs) through Many-Shot Jailbreaking (MSJ). Our experiments utilize context length of up to 128K tokens. Through comprehensive analysis with various many-shot attack settings with different instruction styles, shot density, topic, and format, we reveal that context length is the primary factor determining attack effectiveness. Critically, we find that successful attacks do not require carefully crafted harmful content. Even repetitive shots or random dummy text can circumvent model safety measures, suggesting fundamental limitations in long-context processing capabilities of LLMs. The safety behavior of well-aligned models becomes increasingly inconsistent with longer contexts. These findings highlight significant safety gaps in context expansion capabilities of LLMs, emphasizing the need for new safety mechanisms.

cs.CL↗

arXiv2025

LLM-C3MOD: A Human-LLM Collaborative System for Cross-Cultural Hate Speech Moderation

Content moderation is a global challenge, yet major tech platforms prioritize high-resource languages, leaving low-resource languages with scarce native moderators. Since effective moderation depends on understanding contextual cues, this imbalance increases the risk of improper moderation due to non-native moderators' limited cultural understanding. Through a user study, we identify that non-native moderators struggle with interpreting culturally-specific knowledge, sentiment, and internet culture in the hate speech moderation. To assist them, we present LLM-C3MOD, a human-LLM collaborative pipeline with three steps: (1) RAG-enhanced cultural context annotations; (2) initial LLM-based moderation; and (3) targeted human moderation for cases lacking LLM consensus. Evaluated on a Korean hate speech dataset with Indonesian and German participants, our system achieves 78% accuracy (surpassing GPT-4o's 71% baseline), while reducing human workload by 83.6%. Notably, human moderators excel at nuanced contents where LLMs struggle. Our findings suggest that non-native moderators, when properly supported by LLMs, can effectively contribute to cross-cultural hate speech moderation.

cs.CL↗

arXiv2025

HEISIR: Hierarchical Expansion of Inverted Semantic Indexing for Training-free Retrieval of Conversational Data using LLMs

The growth of conversational AI services has increased demand for effective information retrieval from dialogue data. However, existing methods often face challenges in capturing semantic intent or require extensive labeling and fine-tuning. This paper introduces HEISIR (Hierarchical Expansion of Inverted Semantic Indexing for Retrieval), a novel framework that enhances semantic understanding in conversational data retrieval through optimized data ingestion, eliminating the need for resource-intensive labeling or model adaptation. HEISIR implements a two-step process: (1) Hierarchical Triplets Formulation and (2) Adjunct Augmentation, creating semantic indices consisting of Subject-Verb-Object-Adjunct (SVOA) quadruplets. This structured representation effectively captures the underlying semantic information from dialogue content. HEISIR achieves high retrieval performance while maintaining low latency during the actual retrieval process. Our experimental results demonstrate that HEISIR outperforms fine-tuned models across various embedding types and language models. Beyond improving retrieval capabilities, HEISIR also offers opportunities for intent and topic analysis in conversational data, providing a versatile solution for dialogue systems.

cs.IR↗

arXiv2024

Inorganic electrochromic metasurface in the visible

Colour printing based on metallic or dielectric nanostructures has revolutionized colour science due to its unprecedented subwavelength resolution. Evidently, the evolution towards the active control of such structural colours with smart materials is in progress for real applications. Here we experimentally demonstrate a large colour gamut with high intensity and purity, as well as its switching on and off based solely on tungsten trioxide (WO3) cylindrical resonators. The strong resonances in the visible spectral range in these WO3 metasurfaces can be reversibly switched on and off due to its electrochromism by applying alternating voltages of +2.0 V and -0.3 V. Our approach opens up possibilities for the functional diversification of commercial smart windows, as well as the development of new display technologies in the future.

physics.optics↗

arXiv2024

Polarization-independent metasurfaces based on bound states in the continuum with high Q-factor and resonance modulation

Metasurfaces offer a powerful platform for effective light manipulation, which is crucial for advanced optical technologies. While designs of polarization-independent structures have reduced the need for polarized illumination, they are often limited by either low Q factors or low resonance modulation. Here, we design and experimentally demonstrate a metasurface with polarization-independent quasi-bound state in the continuum (quasi-BIC), where the unit cell consists of four silicon squares arranged in a two-dimensional array and the resonance properties can be controlled by adjusting the edge length difference between different squares. Our metasurface experimentally achieves a Q factor of approximately 100 and a resonance modulation of around 50%. This work addresses a common limitation in previous designs, which either achieved high Q factors exceeding 200 with a resonance modulation of less than 10%, leading to challenging signal-to-noise ratio requirements, or achieved strong resonance modulation with Q factors of only around 10, limiting light confinement and fine-tuning capabilities. In contrast, our metasurface ensures that the polarization-independent signal is sharp and distinct within the system, reducing the demands on signal-to-noise ratio and improving robustness. Experiments show the consistent performance across different polarization angles. This work contributes to the development of versatile optical devices, enhancing the potential for the practical application of BIC-based designs in areas such as optical filtering and sensing.

physics.optics↗

arXiv2024

Harmonizing Attention: Training-free Texture-aware Geometry Transfer

Extracting geometry features from photographic images independently of surface texture and transferring them onto different materials remains a complex challenge. In this study, we introduce Harmonizing Attention, a novel training-free approach that leverages diffusion models for texture-aware geometry transfer. Our method employs a simple yet effective modification of self-attention layers, allowing the model to query information from multiple reference images within these layers. This mechanism is seamlessly integrated into the inversion process as Texture-aligning Attention and into the generation process as Geometry-aligning Attention. This dual-attention approach ensures the effective capture and transfer of material-independent geometry features while maintaining material-specific textural continuity, all without the need for model fine-tuning.

cs.CV↗

arXiv2022

Automated Evaluation for Student Argumentative Writing: A Survey

This paper surveys and organizes research works in an under-studied area, which we call automated evaluation for student argumentative writing. Unlike traditional automated writing evaluation that focuses on holistic essay scoring, this field is more specific: it focuses on evaluating argumentative essays and offers specific feedback, including argumentation structures, argument strength trait score, etc. The focused and detailed evaluation is useful for helping students acquire important argumentation skill. In this paper we organize existing works around tasks, data and methods. We further experiment with BERT on representative datasets, aiming to provide up-to-date baselines for this field.

cs.CL↗