arXiv subjects

Jaewon Lee

Publications and source records attributed to Jaewon Lee.

At least 19 recordsLinked to original sources

arXiv2026

Towards Standardized Evaluation of GPU Memory Safety with GMSBench

As GPUs become increasingly integral to high-performance computing and machine learning, ensuring memory safety in GPU programs has become crucial for reliable and secure execution. However, evaluating GPU memory safety techniques remains challenging due to the lack of comprehensive and standardized benchmarks. In this paper, we present GMSBench, a GPU memory safety benchmark designed to evaluate a broad range of memory safety violations across different GPU memory spaces and execution scenarios. GMSBench comprises 149 self-contained CUDA tests spanning spatial, temporal, and concurrency errors. The suite provides a standardized foundation for the evaluation and comparative analysis of GPU memory safety mechanisms and helps expose gaps in their detection coverage. We demonstrate the utility of GMSBench by evaluating Compute Sanitizer, a widely used GPU memory error detection tool across multiple GPU architectures.

cs.CR↗

arXiv2026

Graph-GSReg: Leveraging 3D Scene Graphs for Gaussian Splatting Registration

Merging multiple 3D Gaussian Splatting (3DGS) scenes into a single unified Gaussian representation is essential for large-scale 3D mapping and long-term map management. Despite its importance, this area remains underexplored, and existing solutions exhibit several limitations. Learning-based methods attempt direct correspondence between Gaussian primitives and require training on large 3DGS datasets. Image-based optimization methods depend heavily on coarse initialization from generic foundation models and often incur expensive refinement. We present \ourmodel. Our method constructs a 3D scene graph from a 3DGS and its rendered images, \textit{reformulating 3DGS registration as a graph registration problem}. The proposed 3D scene graph represents each 3DGS at a higher-level representation, enabling a globally consistent understanding of semantic information and structural context for accurate registration. To further construct a seamless unified scene, we introduce a Self-Supervised Test-Time Optimization. Naively merging two 3D Gaussian scenes often suffers from occlusion artifacts such as hollows and floaters. To alleviate this issue, we refine the merged Gaussians to preserve visual consistency between the original scenes and the merged scene. We evaluate our method on real and synthetic benchmarks, demonstrating competitive registration accuracy and merged scene rendering quality.

cs.CV↗

arXiv2026

KSAFE-MM: A Multimodal Safety Benchmark via Localized Contextualization for Korean Cultural Risks

Multimodal Large Language Models (MLLMs) exacerbate safety risks by introducing vulnerabilities across multiple modalities, such as language and vision. Current MLLM safety evaluation tools, however, suffer from major limitations: 1) English-centric dataset construction, and 2) a focus on generic risks that are not tied to local cultural contexts. This paper introduces KSAFE-MM, a benchmark for Korean multimodal safety evaluation that covers both general safety risks and culture-specific vulnerabilities. KSAFE-MM consists of two parts, KSAFE-MM-G and KSAFE-MM-C. KSAFE-MM-G evaluates globally shared risks in Korean contexts through linguistic contextualization, which transforms generic safety queries into contextually grounded multimodal samples. KSAFE-MM-C targets culture-dependent MLLM safety vulnerabilities using localized visual queries derived from real-world contexts. It pairs these visual queries with jailbreak-style textual queries to cover multimodal safety risks involving cultural visual cues and malicious textual intent. Together, these components provide a general-to-local construction pipeline for evaluating both globally shared safety risks and culture-specific vulnerabilities. We evaluate 12 state-of-the-art MLLMs on KSAFE-MM and reveal that models exhibit greater vulnerability to culturally grounded attacks than to generic ones. Notably, jailbreaking strategies substantially amplify attack success rates, with ProgramExecution yielding up to 74.2% ASR compared to 13.4% for standard queries. Furthermore, we identify a systematic trade-off between safety and over-refusal, where models achieving low ASR tend to exhibit excessive refusal behavior on benign queries. These findings highlight the urgent need for culturally grounded safety evaluation beyond English-centric benchmarks.

cs.CL↗

arXiv2026

Instruct-FD: Can Your Full-Duplex Speech System Follow Turn-Taking Instructions?

Current full-duplex (FD) spoken dialogue systems can produce fluid interactions, yet it remains unclear whether they can adapt their turn-taking behavior when explicitly instructed. This is critical for real-world deployment, where conversational policies vary across applications (e.g., proactive tutoring vs. passive counseling). We introduce Instruct-FD, an instruction-conditioned benchmark for evaluating controllable turn management in FD systems. To enable this, we develop a human-validated, scalable synthetic pipeline that generates instruction-conditioned conversations, along with a deployment-agnostic multi-turn evaluation protocol and an LLM-based judge. Benchmarking six state-of-the-art full-duplex systems reveals a substantial gap in instruction-following turn management: the best model achieves only 64.4% adherence. Performance is highly uneven across behaviors and scenarios, with proactive behaviors such as model backchanneling and interruption remaining particularly challenging. These findings establish instruction-following turn management as a crucial direction for building adaptable and deployable full-duplex dialogue systems.

cs.CL↗

arXiv2026

Uncertainty Estimation via Hyperspherical Confidence Mapping

Quantifying uncertainty in neural network predictions is essential for high-stakes domains such as autonomous driving, healthcare, and manufacturing. While existing approaches often depend on costly sampling or restrictive distributional assumptions, we propose Hyperspherical Confidence Mapping (HCM), a simple yet principled framework for sampling-free and distribution-free uncertainty estimation. HCM decomposes outputs into a magnitude and a normalized direction vector constrained to lie on the unit hypersphere, enabling a novel interpretation of uncertainty as the degree of violation of this geometric constraint. This yields deterministic and interpretable estimates applicable to both regression and classification. Experiments across diverse benchmarks and real-world industrial tasks demonstrate that HCM matches or surpasses ensemble and evidential approaches, with far lower inference cost and stronger confidence-error alignment. Our results highlight the power of geometric structure in uncertainty estimation and position HCM as a versatile alternative to conventional techniques.

cs.LG↗

arXiv2026

Scaling Multi-Node Mixture-of-Experts Inference Using Expert Activation Patterns

Most recent state-of-the-art (SOTA) large language models (LLMs) use Mixture-of-Experts (MoE) architectures to scale model capacity without proportional per-token compute, enabling higher-quality outputs at manageable serving costs. However, MoE inference at scale is fundamentally bottlenecked by expert load imbalance and inefficient token routing, especially in multi-node deployments where tokens are not guaranteed to be routed to local experts, resulting in significant inter-node all-to-all communication overhead. To systematically characterize these challenges, we profile SOTA open-source MoE models, including Llama 4 Maverick, DeepSeek V3-671B, and Qwen3-230B-A22B, on various datasets and collected over 100k real expert activation traces. Upon studying the expert activation patterns, we uncover various persistent properties across all the frontier MoE models: variable expert load imbalance, domain-specific expert activation where expert popularity shifts across task families (code, math, chat, general), and a strong correlation between prefill and decode expert activations. Motivated by these findings, we propose workload-aware micro-batch grouping and an expert placement strategy to maximize token locality to the destination expert, thereby reducing inter-node communication. Across models and datasets, these optimizations help reduce all2all communication data up to 20, resulting in lower MoE decode latency and better accelerator utilization.

cs.LG↗

arXiv2026

Byte-level generative predictions for forensics multimedia carving

Digital forensic investigations often face significant challenges when recovering fragmented multimedia files that lack file system metadata. While traditional file carving relies on signatures and discriminative deep learning models for fragment classification, these methods cannot reconstruct or predict missing data. We propose a generative approach to multimedia carving using bGPT, a byte-level transformer designed for next-byte prediction. By feeding partial BMP image data into the model, we simulate the generation of likely fragment continuations. We evaluate the fidelity of these predictions using different metrics, namely, cosine similarity, structural similarity index (SSIM), chi-square distance, and Jensen-Shannon divergence (JSD). Our findings demonstrate that generative models can effectively predict byte-level patterns to support fragment matching in unallocated disk space.

cs.CV↗

arXiv2026

KMTNet Synoptic Survey of Southern Sky III: The First Data Release

We present the first public data release (DR1) of the KMTNet Synoptic Survey of Southern Sky (KS4). This deep, wide-field imaging survey covers a southern footprint of -85$^{\circ}$ < Decl. < -28.8$^{\circ}$ in the $B$, $V$, $R$, and $I$ bands using a network of three 1.6-m telescopes. Although primarily designed to secure reference imaging for gravitational wave counterpart identification, DR1 delivers science-ready data for $\sim$4,000 deg$^{2}$ to enable a broad range of astrophysical research. The release includes deep co-added images reaching median 5$\sigma$ depths of 22.0-23.5 AB mag. It is accompanied by two source catalogs containing over 200 million sources with SNR $>5$: an $I$-band-selected forced-photometry catalog optimized for consistent colors, and a band-merged catalog offering enhanced completeness. Validation demonstrates robust data quality, characterized by mean astrometric offsets of $+0.054 \pm 0.129$ arcsec in RA and $-0.015 \pm 0.120$ arcsec in Dec relative to Gaia DR3. {\refbf Photometric uniformity for point sources is maintained within $\pm 0.03$ mag relative to Gaia XP for 97.5--99.8\% of the footprint across all four bands.} A key advantage of KS4 is its uniform and contiguous spatial coverage. It extends to fainter magnitudes than other uniform surveys while filling irregular gaps in existing deep datasets. All data products are publicly available via the CDS and NOIRLab's Astro Data Lab.

astro-ph.IM↗

arXiv2026

Back to Basics: Revisiting ASR in the Age of Voice Agents

Automatic speech recognition (ASR) systems have achieved near-human accuracy on curated benchmarks, yet still fail in real-world voice agents under conditions that current evaluations do not systematically cover. Without diagnostic tools that isolate specific failure factors, practitioners cannot anticipate which conditions, in which languages, will cause what degree of degradation. We introduce WildASR, a multilingual (four-language) diagnostic benchmark sourced entirely from real human speech that factorizes ASR robustness along three axes: environmental degradation, demographic shift, and linguistic diversity. Evaluating seven widely used ASR systems, we find severe and uneven performance degradation, and model robustness does not transfer across languages or conditions. Critically, models often hallucinate plausible but unspoken content under partial or degraded inputs, creating concrete safety risks for downstream agent behavior. Our results demonstrate that targeted, factor-isolated evaluation is essential for understanding and improving ASR reliability in production systems. Besides the benchmark itself, we also present three analytical tools that practitioners can use to guide deployment decisions.

cs.AI↗

arXiv2026

DALDALL: Data Augmentation for Lexical and Semantic Diverse in Legal Domain by leveraging LLM-Persona

Data scarcity remains a persistent challenge in low-resource domains. While existing data augmentation methods leverage the generative capabilities of large language models (LLMs) to produce large volumes of synthetic data, these approaches often prioritize quantity over quality and lack domain-specific strategies. In this work, we introduce DALDALL, a persona-based data augmentation framework tailored for legal information retrieval (IR). Our method employs domain-specific professional personas--such as attorneys, prosecutors, and judges--to generate synthetic queries that exhibit substantially greater lexical and semantic diversity than vanilla prompting approaches. Experiments on the CLERC and COLIEE benchmarks demonstrate that persona-based augmentation achieves improvement in lexical diversity as measured by Self-BLEU scores, while preserving semantic fidelity to the original queries. Furthermore, dense retrievers fine-tuned on persona-augmented data consistently achieve competitive or superior recall performance compared to those trained on original data or generic augmentations. These findings establish persona-based prompting as an effective strategy for generating high-quality training data in specialized, low-resource domains.

cs.CL↗

arXiv2026

KMTNet Synoptic Survey of Southern Sky II: Data Reduction and Real-Time Transient Detection Pipeline

We present a comprehensive pipeline developed for the image processing of the KMTNet Synoptic Survey of the Southern Sky (KS4) Data Release 1. This pipeline encompasses several key processes, including data quality assurance, astrometry, photometric zero-point (ZP) calibration, bad pixel masking, image stacking, and difference image analysis (DIA). The astrometric solutions were validated by cross-matching with the Gaia EDR3 catalog, achieving sub-pixel astrometric accuracy (< 0.4 arcsec). To ensure spatial consistency, we divided each image into multiple subsections and confirmed that astrometric accuracy was maintained even at the edges. We performed a two-stage photometric calibration. Initial ZP solutions were computed for each individual image frame using the APASS DR9 and SkyMapper DR3 catalogs. Subsequently, we corrected residual spatial variations in the stacked images using Gaia XP photometry. This procedure yielded a 5-sigma depth of 22-23 AB mag across the BVRI bands, with root-mean-square errors of approximately 0.03 mag when referenced to Gaia stars in the magnitude range of 14-19 mag. The processed KS4 images span over 4,000 deg^2 of the southern sky, providing reference images suitable for DIA. This publicly available pipeline also supports real-time processing of newly acquired images, enabling prompt transient detection. We demonstrate its effectiveness through successful applications in gravitational-wave follow-up observations.

astro-ph.IM↗

arXiv2026

Rethinking Pose Refinement in 3D Gaussian Splatting under Pose Prior and Geometric Uncertainty

3D Gaussian Splatting (3DGS) has recently emerged as a powerful scene representation and is increasingly used for visual localization and pose refinement. However, despite its high-quality differentiable rendering, the robustness of 3DGS-based pose refinement remains highly sensitive to both the initial camera pose and the reconstructed geometry. In this work, we take a closer look at these limitations and identify two major sources of uncertainty: (i) pose prior uncertainty, which often arises from regression or retrieval models that output a single deterministic estimate, and (ii) geometric uncertainty, caused by imperfections in the 3DGS reconstruction that propagate errors into PnP solvers. Such uncertainties can distort reprojection geometry and destabilize optimization, even when the rendered appearance still looks plausible. To address these uncertainties, we introduce a relocalization framework that combines Monte Carlo pose sampling with Fisher Information-based PnP optimization. Our method explicitly accounts for both pose and geometric uncertainty and requires no retraining or additional supervision. Across diverse indoor and outdoor benchmarks, our approach consistently improves localization accuracy and significantly increases stability under pose and depth noise.

cs.CV↗

arXiv2026

RVN-Bench: A Benchmark for Reactive Visual Navigation

Safe visual navigation is critical for indoor mobile robots operating in cluttered environments. Existing benchmarks, however, often neglect collisions or are designed for outdoor scenarios, making them unsuitable for indoor visual navigation. To address this limitation, we introduce the reactive visual navigation benchmark (RVN-Bench), a collision-aware benchmark for indoor mobile robots. In RVN-Bench, an agent must reach sequential goal positions in previously unseen environments using only visual observations and no prior map, while avoiding collisions. Built on the Habitat 2.0 simulator and leveraging high-fidelity HM3D scenes, RVN-Bench provides large-scale, diverse indoor environments, defines a collision-aware navigation task and evaluation metrics, and offers tools for standardized training and benchmarking. RVN-Bench supports both online and offline learning by offering an environment for online reinforcement learning, a trajectory image dataset generator, and tools for producing negative trajectory image datasets that capture collision events. Evaluations demonstrate that policies trained on RVN-Bench generalize effectively across unseen simulated environments. Furthermore, initial physical experiments using a Jackal UGV indicate promising sim-to-real transfer. Code and additional materials are available at: https://sequor-robotics-research.github.io/projects/RVN-Bench/.

cs.RO↗

arXiv2026

Radiological mapping and uncertainty quantification by a fast Microcanonical Langevin Monte Carlo sampler

Radiological mapping plays a critical role in nuclear emergency response and environmental management activities. A radiation image, representing the spatial and intensity distribution of the radioactivity, is reconstructed from the radiation data and the associated contextual information. Typical image reconstruction methods, such as Maximum Likelihood Expectation-Maximization (ML-EM), only provide point estimates of the pixel or voxel activities without associated uncertainties. Here, we apply a new Microcanonical Langevin Monte Carlo (MCLMC) sampler for radiation image reconstruction and uncertainty quantification. The MCLMC sampler properties are first tested with synthetic radiation images. Methods to obtain the radiation distribution estimate and the associated uncertainty from the samples drawn by MCLMC are discussed. Given sufficient measurement statistics, the radiation distribution estimate obtained from MCLMC results closely resembles the ground truth with less risk of over- or under-fitting compared to ML-EM. When MCLMC is run in parallel on a GPU, the samples can converge to the posterior distribution in about 10 seconds for an image with $10^3$--$10^4$ pixels, which is significantly faster than other comparable Markov Chain Monte Carlo (MCMC) samplers. We also tested MCLMC on a dataset from a real distributed source radiological mapping campaign, and the reconstructed results agree well with ground truth. The fast MCLMC sampler therefore enables improved imaging accuracy and prompt uncertainty quantification for reconstructed radiation images, which can better inform decision-making in response to radiological events.

physics.med-ph↗

arXiv2026

Consistency-Preserving Concept Erasure via Unsafe-Safe Pairing and Directional Fisher-weighted Adaptation

With the increasing versatility of text-to-image diffusion models, the ability to selectively erase undesirable concepts (e.g., harmful content) has become indispensable. However, existing concept erasure approaches primarily focus on removing unsafe concepts without providing guidance toward corresponding safe alternatives, which often leads to failure in preserving the structural and semantic consistency between the original and erased generations. In this paper, we propose a novel framework, PAIRed Erasing (PAIR), which reframes concept erasure from simple removal to consistency-preserving semantic realignment using unsafe-safe pairs. We first generate safe counterparts from unsafe inputs while preserving structural and semantic fidelity, forming paired unsafe-safe multimodal data. Leveraging these pairs, we introduce two key components: (1) Paired Semantic Realignment, a guided objective that uses unsafe-safe pairs to explicitly map target concepts to semantically aligned safe anchors; and (2) Fisher-weighted Initialization for DoRA, which initializes parameter-efficient low-rank adaptation matrices using unsafe-safe pairs, encouraging the generation of safe alternatives while selectively suppressing unsafe concepts. Together, these components enable fine-grained erasure that removes only the targeted concepts while maintaining overall semantic consistency. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art baselines, achieving effective concept erasure while preserving structural integrity, semantic coherence, and generation quality.

cs.CV↗

arXiv2026

Erase at the Core: Representation Unlearning for Machine Unlearning

Many approximate machine unlearning methods demonstrate strong logit-level forgetting -- such as near-zero accuracy on the forget set -- yet continue to preserve substantial information within their internal feature representations. We refer to this discrepancy as superficial forgetting. Recent studies indicate that most existing unlearning approaches primarily alter the final classifier, leaving intermediate representations largely unchanged and highly similar to those of the original model. To address this limitation, we introduce the Erase at the Core (EC), a framework designed to enforce forgetting throughout the entire network hierarchy. EC integrates multi-layer contrastive unlearning on the forget set with retain set preservation through deeply supervised learning. Concretely, EC attaches auxiliary modules to intermediate layers and applies both contrastive unlearning and cross-entropy losses at each supervision point, with layer-wise weighted losses. Experimental results show that EC not only achieves effective logit-level forgetting, but also substantially reduces representational similarity to the original model across intermediate layers. Furthermore, EC is model-agnostic and can be incorporated as a plug-in module into existing unlearning methods, improving representation-level forgetting while maintaining performance on the retain set.

cs.LG↗

arXiv2026

Forecasting Future Language: Context Design for Mention Markets

Mention markets, a type of prediction market in which contracts resolve based on whether a specified keyword is mentioned during a future public event, require accurate probabilistic forecasts of keyword-mention outcomes. While recent work shows that large language models (LLMs) can generate forecasts competitive with human forecasters, it remains unclear how input context should be designed to support accurate prediction. In this paper, we study this question through experiments on earnings-call mention markets, which require forecasting whether a company will mention a specified keyword during its upcoming call. We run controlled comparisons varying (i) which contextual information is provided (news and/or prior earnings-call transcripts) and (ii) how \textit{market probability}, (i.e., prediction market contract price) is used. We introduce Market-Conditioned Prompting (MCP), which explicitly treats the market-implied probability as a prior and instructs the LLM to update this prior using textual evidence, rather than re-predicting the base rate from scratch. In our experiments, we find three insights: (1) richer context consistently improves forecasting performance; (2) market-conditioned prompting (MCP), which treats the market probability as a prior and updates it using textual evidence, yields better-calibrated forecasts; and (3) a mixture of the market probability and MCP (MixMCP) outperforms the market baseline. By dampening the LLM's posterior update with the market prior, MixMCP yields more robust predictions than either the market or the LLM alone.

q-fin.GN↗

arXiv2026

CuFuzz: Hardening CUDA Programs through Transformation and Fuzzing

GPUs have gained significant popularity over the past decade, extending beyond their original role in graphics rendering. This evolution has brought GPU security and reliability to the forefront of concerns. Prior research has shown that CUDA's lack of memory safety can lead to serious vulnerabilities. While fuzzing is effective for finding such bugs on CPUs, equivalent tools for GPUs are lacking due to architectural differences and lack of built-in error detection. In this paper, we propose CuFuzz, a novel compiler-runtime co-design solution to extend state-of-the-art CPU fuzzing tools to GPU programs. CuFuzz transforms GPU programs into CPU programs using compiler IR-level transformations to enable effective fuzz testing. To the best of our knowledge, CuFuzz is the first mechanism to bring fuzzing support to CUDA, addressing a critical gap in GPU security research. By leveraging CPU memory error detectors such as Address Sanitizer, CuFuzz aims to uncover memory safety bugs and related correctness vulnerabilities in CUDA code, enhancing the security and reliability of GPU-accelerated applications. To ensure high fuzzing throughput, we introduce two compiler-runtime co-optimizations tailored for GPU code: Partial Representative Execution (PREX) and Access-Index Preserving Pruning (AXIPrune), achieving average throughput improvements of 32x with PREX and an additional 33% gain with AXIPrune on top of PREX-optimized code. Together, these optimizations can yield up to a 224.31x speedup. In our fuzzing campaigns, CuFuzz uncovered 122 security vulnerabilities in widely used benchmarks.

cs.CR↗