arXiv subjects

Jaemin Kim

Publications and source records attributed to Jaemin Kim.

At least 19 recordsLinked to original sources

arXiv2026

A priori error estimator for reduced-order models based on the higher-order Craig-Bampton method in dynamic substructuring

The Craig-Bampton (CB) method is a widely used dynamic substructuring technique based on component mode synthesis (CMS). The higher-order Craig-Bampton (HCB) method augments the CB basis with residual modes from a Neumann series expansion of the residual flexibility matrix, where HCB-n retains terms up to the n-th order and is reduced back to the CB size via the System Equivalent Reduction Expansion Process (SEREP), achieving improved accuracy at the same model dimension. However, assessing the accuracy of a reduced model without solving the full-order problem remains a fundamental challenge: if the full-order solution is required to evaluate the error, the purpose of model reduction is defeated. Despite the demonstrated superiority of the HCB method, no a priori error estimator (one that predicts eigenvalue errors without solving the full-order eigenvalue problem) has been proposed for it. The present work addresses this gap with a hierarchical estimation framework that exploits the nested Ritz subspace structure of the HCB method, where each higher-order solution serves as a reference for estimating the error of the preceding order. The framework provides (i) a generalized CB error estimator derived from a Rayleigh quotient perturbation analysis, and (ii) a novel HCB-1 error estimator using the HCB-2 eigensolution as a reference. Numerical examples across models of varying geometric complexity validate both estimators.

math.NA↗

arXiv2026

A physics-informed neural network for improving surface reconstruction of intracranial saccular aneurysms via variational membrane equilibrium

Intracranial saccular aneurysms (ISAs) pose severe health risks, yet conventional population-based risk stratification scores (PHASES, UIATS, and ELAPSS) offer limited capacity for patient-specific rupture risk assessment. Image-based computational approaches have gained prominence, but traditional surface reconstruction relies on mathematical smoothing (e.g., L-curve criteria) that indiscriminately suppresses both imaging artifacts and genuine pathological features such as rupture-prone blebs. Although Laplace's membrane equilibrium ($\kappa_1 T_1 + \kappa_2 T_2 = P$) has long governed aneurysm wall mechanics (Humphrey and Kyriacou [Neurol. Res., 18 (1996)]), its integration into geometric reconstruction pipelines remains unexplored. This work introduces a physics-informed neural network (PINN) framework with B-spline representations, whose key contributions are: (i) embedding the variational equilibrium condition ($\delta \Pi = 0$) as a physics-informed loss that replaces the mathematical L-curve criterion with a biomechanically grounded artifact discrimination, (ii) developing a Manifold-Consistent CNN ansatz that preserves the closed-surface topology of vascular geometries, and (iii) establishing a Laplace equilibrium-driven reconstruction that filters imaging noise while preserving diagnostically critical high-curvature features. Application to patient-specific clinical datasets demonstrates that the framework eliminates non-physical concave artifacts without compromising genuine geometric anomalies. Clinical evaluation by a practicing neurosurgeon confirms that the resulting risk map---with rupture risk concentrated at the dome apex---is consistent with intraoperative observations, establishing a computational biomarker foundation for patient-specific rupture risk assessment of intracranial saccular aneurysms.

math.NA↗

arXiv2026

Ciphertext- and Polynomial-Level Optimization for Fully Homomorphic Encryption

Fully homomorphic encryption (FHE) schemes such as RNS-CKKS enable privacy-preserving services through direct computation on encrypted data. While recent FHE compilers optimize FHE programs, they operate at the coarse-grained ciphertext level, where each ciphertext operation comprises a sequence of polynomial operations. At this granularity, the compilers miss polynomial-level optimization opportunities across ciphertext operations. This work presents Recifhe, a new multi-level compiler that supports both ciphertext-level and polynomial-level optimization. At the ciphertext level, Recifhe transforms a non-FHE input program into an FHE program by inserting ciphertext management operations and applies global optimizations. At the polynomial level, Recifhe eliminates redundant polynomial computations across ciphertext operations. Recifhe achieves a 1.25x speedup over ciphertext-level-only optimization.

cs.CR↗

arXiv2026

PDE-constrained optimization for virtual sensing in structural dynamics: Full-field displacement and force recovery from sparse sensors

Virtual sensing -- recovering full-field structural response from sparse sensor measurements -- is a fundamental challenge in structural health monitoring (SHM). Fracture assessment, fatigue evaluation and remaining useful life (RUL) prognosis all take the applied load as their input, and that load frequently acts on surfaces that cannot be instrumented. An improved load estimate is therefore of direct practical value for SHM and for prognostics and health management (PHM). The mode-based approach reconstructs the displacement field accurately from a small number of sensors and has been applied successfully for that purpose. The load inferred from this reconstructed displacement is far less reliable, because the residual displacement error is amplified when it is mapped back through the dynamic stiffness. This study therefore formulates virtual sensing as a PDE-constrained optimization (PDE-CO) problem in which the elastodynamic equation is an equality constraint and the applied load is the optimization variable, so that displacement and load are estimated together rather than in sequence. To manage the computational cost, the framework separates offline finite element model preparation from online reconstruction. The assembled matrices are reused during online optimization, and the forward and gradient evaluations can use GPU-based parallel computation. The numerical comparisons support improved noise robustness of PDE-CO displacement reconstruction relative to the mode-based approach within the tested settings. These results provide a theoretical and computational foundation for the development of virtual sensing for SHM and PHM.

math-ph↗

arXiv2026

Coupling-Robust Accuracy in Multiphysics Physics Informed Neural Networks via Kronecker-Preconditioned Optimization

Physics-informed neural networks (PINNs) offer a mesh-free route to solving coupled multiphysics systems, but their accuracy degrades systematically as inter-equation coupling strengthens, and inverse-gradient-norm loss balancing alone does not reliably prevent this failure. This study explains why coupling degrades PINN training and identifies an optimizer structure that removes the dependence, replacing case-by-case tuning with a principled remedy. Through a neural tangent kernel analysis, we prove that the standard kernel's spectral radius grows as $\Omega(\gamma^2)$ with coupling strength $\gamma$, whereas block-diagonal Gauss-Newton (GN) preconditioning bounds it by the number of networks $S$, independent of $\gamma$; no diagonal preconditioner recovers this bound for any coupling type or loss weighting. We realize block-diagonal GN preconditioning through the Kronecker-preconditioned optimizer SOAP combined with inverse-gradient-norm loss balancing (SOAP+GradNorm) and evaluate it across 222 experiments on four benchmarks of increasing difficulty. Across all systems, SOAP+GradNorm is the only configuration whose degradation remains bounded in every regime tested: it preserves weak-coupling accuracy in linear problems and limits degradation to $2.3\times$ in the nonlinear Nernst-Planck-Poisson system, whereas Adam-based training leaves the $L_2$ error above the 0.1 failure threshold. The same effect applies to a six-residual, four-network, 2D electro-osmotic flow where the electric double layer is resolved down to a Debye length of $\varepsilon = 0.01$ on an $x$-invariant reference solution. These results recast coupling-induced accuracy loss as a problem of the preconditioner's structure rather than loss weighting and identify Kronecker preconditioning as a structural lever for training PINNs on strongly coupled, stiff multiphysics systems.

cs.LG↗

arXiv2026

From Volume to Value: Preference-Aligned Memory Construction for On-Device RAG

With the rapid emergence of personal AI agents based on Large Language Models (LLMs), implementing them on-device has become essential for privacy and responsiveness. To handle the inherently personal and context-dependent nature of real-world requests, such agents must ground their generation in device-resident personal context. However, under tight memory budgets, the core bottleneck is what to store so that retrieval remains aligned with the user. We propose EPIC (Efficient Preference-aligned Index Construction), which focuses on user preferences as a compact and stable form of personal context and integrates them throughout the RAG pipeline. EPIC selectively retains preference-relevant information from raw data and aligns retrieval toward preference-aligned contexts. Across four benchmarks covering conversations, debates, explanations, and recommendations, EPIC reduces indexing memory by 2,404 times, improves preference-following accuracy by 18.79 %p, and achieves 32.17 times lower retrieval latency over the best-performing baseline. In on-device experiments, EPIC maintains under 1 MB memory and achieves 5.21 to 29.35 ms/query latency across three platforms, while supporting streaming updates under preference drift. Our code and data are available at https://github.com/UbiquitousAILab/EPIC.

cs.CL↗

arXiv2026

AgentPSO: Evolving Agent Reasoning Skill via Multi-agent Particle Swarm Optimization

Multi-agent reasoning has shown promise for improving the problem-solving ability of large language models by allowing multiple agents to explore diverse reasoning paths. However, most existing multi-agent methods rely on inference-time debate or aggregation, which can be vulnerable to incorrect peer influence and biased consensus. Moreover, the agents' reasoning procedures remain static, as they do not accumulate reusable behaviors across tasks. In this paper, we introduce \textbf{AgentPSO}, a particle-swarm-inspired framework for evolving multi-agent reasoning skills. AgentPSO treats each agent as a particle-like reasoner whose state is a natural-language skill and whose velocity is a semantic update direction, iteratively guiding agents toward higher-performing skill configurations. Across training iterations, each agent updates its skill by combining its previous velocity, personal-best skill, global-best skill, and a self-reflective direction derived from peer reasoning trajectories. This enables agents to learn reusable reasoning behaviors by drawing on their own experience and on the strongest skills found by the population, without updating the parameters of the backbone language model. Experiments on mathematical and general reasoning benchmarks show that AgentPSO improves over static single-agent skills and test-time-only multi-agent reasoning baselines. The evolved skills further transfer across benchmarks and to another backbone model, suggesting that AgentPSO captures reusable reasoning procedures rather than merely optimizing benchmark-specific prompts. Code is publicly available at https://github.com/HYUNMIN-HWANG/AgentPSO/.

cs.AI↗

arXiv2026

On the complementary roles of anisotropic crack density and anisotropic crack driving force in phase-field modeling of mixed-mode fracture

Phase-field models for anisotropic fracture employ two complementary mechanisms: (i) the anisotropic crack density function, controlling direction-dependent fracture resistance, and (ii) the anisotropic strain energy, governing the fracture driving force. Although the unified framework was presented in Pranavi et al.[Comput. Mech., 73 (2024)], the distinct roles of these mechanisms and their interaction remain uninvestigated. This work addresses this gap by first validating the formulation against mixed-mode fracture experiments on a soft elastomer (Lu et al. [Extreme Mech. Lett., 48 (2021)]), and then conducting systematic parametric studies on single-edge-notched (SEN) and open-hole tension (OHT) specimens to isolate each mechanism. The SEN studies show that the crack density anisotropy controls the crack path and toughness while leaving the elastic response unchanged, whereas the anisotropic strain energy deflects the crack but saturates rapidly. The OHT studies reveal a geometry-dependent role expansion: the anisotropic strain energy governs fiber-orientation-dependent stiffness, peak force, and fracture displacement. When both mechanisms act together, the combined response exhibits nonlinear synergistic interaction exceeding the linear sum of the individual contributions. These results establish that the crack density anisotropy governs the crack path (fracture resistance), while the anisotropic strain energy governs the driving force and, in stress-concentration geometries, additionally controls the elastic strain energy distribution around the stress concentrator.

cond-mat.mtrl-sci↗

arXiv2026

Robust Ultra Low-Bit Post-Training Quantization via Stable Diagonal Curvature Estimate

Large Language Models (LLMs) are widely used across many domains, but their scale makes deployment challenging. Post-Training Quantization (PTQ) reduces memory footprint without retraining by leveraging a small calibration set. Recent Hessian-based PTQ methods compensate quantization error via cross-channel dependencies, but such approaches degrade at low bit-widths due to noisy curvature estimates from limited calibration data. We propose DASH-Q, a robust PTQ framework using diagonal Hessian approximation and iterative weighted least squares. By discarding noise-prone dependencies, DASH-Q filters sampling noise while prioritizing the preservation of salient feature power. We outperform other PTQ baselines in ultra low-bit regime, improving zero-shot accuracy by 7.01% on average and up to 14.01% over the strongest baselines across five baseline LLM models, while showing robust and stable performance with very small calibration data.

cs.LG↗

arXiv2026

Where Does Robustness Live? Neuron-Guided Adaptation for Retrieval-Augmented Language Models

Retrieval-Augmented Language Models (RALMs) have shown strong potential in knowledge-intensive tasks, yet they remain vulnerable when retrieved contexts are noisy or irrelevant. Robustness against such contexts requires two distinct capabilities: abstention when contexts are uninformative, and selective extraction when relevant evidence is buried in noise. Yet existing methods face two key limitations: they do not train separately for these two capabilities, and they adapt the model at a coarse layer- or module-level granularity, overlooking that only a small subset of neurons is strongly activated for a given input. We propose NeuRIT, a Neuron-guided Robust Instruction-Tuning framework built on a localization-first perspective. NeuRIT mines context-aware neurons associated with relevant and irrelevant context processing, and uses them as anchors to selectively adapt both the identified neuron groups and the layers in which they concentrate. NeuRIT then performs two-stage instruction tuning that teaches complementary behaviors: suppress generation when there is nothing to extract, and extract relevant evidence when there is. NeuRIT consistently outperforms strong baselines across diverse QA benchmarks and generator backbones. Our code is available at https://github.com/HYU-ARK-Lab/NeuRIT.

cs.CL↗

arXiv2026

When Sinks Help or Hurt: Unified Framework for Attention Sink in Large Vision-Language Models

Attention sinks are defined as tokens that attract disproportionate attention. While these have been studied in single modality transformers, their cross-modal impact in Large Vision-Language Models (LVLM) remains largely unexplored: are they redundant artifacts or essential global priors? This paper first categorizes visual sinks into two distinct categories: ViT-emerged sinks (V-sinks), which propagate from the vision encoder, and LLM-emerged sinks (L-sinks), which arise within deep LLM layers. Based on the new definition, our analysis reveals a fundamental performance trade-off: while sinks effectively encode global scene-level priors, their dominance can suppress the fine-grained visual evidence required for local perception. Furthermore, we identify specific functional layers where modulating these sinks most significantly impacts downstream performance. To leverage these insights, we propose Layer-wise Sink Gating (LSG), a lightweight, plug-and-play module that dynamically scales the attention contributions of V-sink and the rest visual tokens. LSG is trained via standard next-token prediction, requiring no task-specific supervision while keeping the LVLM backbone frozen. In most layers, LSG yields improvements on representative multimodal benchmarks, effectively balancing global reasoning and precise local evidence.

cs.CV↗

arXiv2026

CLaD: Planning with Grounded Foresight via Cross-Modal Latent Dynamics

Robotic manipulation involves kinematic and semantic transitions that are inherently coupled via underlying actions. However, existing approaches plan within either semantic or latent space without explicitly aligning these cross-modal transitions. To address this, we propose CLaD, a framework that models how proprioceptive and semantic states jointly evolve under actions through asymmetric cross-attention that allows kinematic transitions to query semantic ones. CLaD predicts grounded latent foresights via self-supervised objectives with EMA target encoders and auxiliary reconstruction losses, preventing representation collapse while anchoring predictions to observable states. Predicted foresights are modulated with observations to condition a diffusion policy for action generation. On LIBERO-LONG benchmark, CLaD achieves 94.7\% success rate, competitive with large VLAs with significantly fewer parameters.

cs.RO↗

arXiv2026

Adaptive Guidance for Retrieval-Augmented Masked Diffusion Models

Retrieval-Augmented Generation (RAG) improves factual grounding by incorporating external knowledge into language model generation. However, when retrieved context is noisy, unreliable, or inconsistent with the model's parametric knowledge, it introduces retrieval-prior conflicts that can degrade generation quality. While this problem has been studied in autoregressive language models, it remains largely unexplored in diffusion-based language models, where the iterative denoising process introduces unique challenges for integrating retrieved context. In this work, we propose Adaptive Retrieval-Augmented Masked Diffusion (ARAM), a training-free adaptive guidance framework for Masked Diffusion Models (MDMs) in RAG settings. ARAM dynamically calibrates the guidance scale during denoising according to the Signal-to-Noise Ratio (SNR) of the distributional shift induced by retrieved context. Intuitively, the model strengthens guidance when the retrieved context provides reliable corrective evidence and suppresses it when the contextual signal is noisy or non-supportive. Extensive experiments on multiple knowledge-intensive QA benchmarks show that ARAM improves overall QA performance over competitive RAG baselines.

cs.CL↗

arXiv2026

DMD-augmented Unpaired Neural Schr\"odinger Bridge for Ultra-Low Field MRI Enhancement

Ultra Low Field (64 mT) brain MRI improves accessibility but suffers from reduced image quality compared to 3 T. As paired 64 mT - 3 T scans are scarce, we propose an unpaired 64 mT $\rightarrow$ 3 T translation framework that enhances realism while preserving anatomy. Our method builds upon the Unpaired Neural Schr\"odinge Bridge (UNSB) with multi-step refinement. To strengthen target distribution alignment, we augment the adversarial objective with DMD2-style diffusion-guided distribution matching using a frozen 3T diffusion teacher. To explicitly constrain global structure beyond patch-level correspondence, we combine PatchNCE with an Anatomical Structure Preservation (ASP) regularizer that enforces soft foreground background consistency and boundary aware constraints. Evaluated on two disjoint cohorts, the proposed framework achieves an improved realism structure trade-off, enhancing distribution level realism on unpaired benchmarks while increasing structural fidelity on the paired cohort compared to unpaired baselines.

cs.CV↗

arXiv2026

A unified variational framework for phase-field fracture and third-medium contact in finite deformation hyperelasticity

This paper presents a unified variational framework that integrates phase-field fracture (PFF) and third-medium contact (TMC) within finite deformation hyperelasticity. The key idea is that both crack and contact are treated through regularization: the sharp crack topology is regularized into a diffuse damage field, while the discrete contact interface is regularized by a compliant fictitious medium with auxiliary fields. This strategy eliminates the need for explicit contact detection or crack tracking algorithms. The framework is validated through two-dimensional three-point bending and three-dimensional Brazilian disk test simulations, demonstrating the interplay between contact-induced stress concentration and crack nucleation/propagation. In particular, the Brazilian disk simulation naturally reproduces secondary crushing-type fracture zones near the contact regions -- a phenomenon consistently observed in experiments yet inaccessible to simplified loading models. These results pave the way for predictive simulation of coupled contact-fracture phenomena without recourse to explicit interface tracking.

physics.comp-ph↗

arXiv2026

Entanglement-swapping measurements for deterministic entanglement distribution

Entanglement swapping is a key primitive for distributing entanglement over quantum networks, but different measurement outcomes can produce end-to-end states with different entanglement, requiring branch-dependent processing or the rejection of unfavorable outcomes. We characterize all projective swapping measurements with full-Schmidt-rank vectors such that, for every pair of pure input links, all outcomes yield the same end-to-end state up to local-unitary corrections. Within this family, the measurements that maximize the average G-concurrence for every input pair are built from complex Hadamard operators, and every outcome individually attains the optimum. Classifying the underlying complex Hadamard operators that preserve optimal deterministic swapping gives one class for $d=2,3$, exactly $72$ classes for $d=5$, and uncountably many whenever $d=4k$. We show further that for $d=2,3$, the corrected end-to-end state in a swapping chain is independent of the swapping order, and discuss noise robustness under depolarizing noise and arbitrary convex input contamination. For pure inputs, these schemes retain every outcome while achieving optimal G-concurrence and therefore eliminate outcome-based postselection.

quant-ph↗

arXiv2026

Dementia-R1: Reinforced Pretraining and Reasoning from Unstructured Clinical Notes for Real-World Dementia Prognosis

While Large Language Models (LLMs) have shown strong performance on clinical text understanding, they struggle with longitudinal prediction tasks such as dementia prognosis, which require reasoning over complex, non-monotonic symptom trajectories across multiple visits. Standard supervised training lacks explicit annotations for symptom evolution, while direct Reinforcement Learning (RL) is hindered by sparse binary rewards. To address this challenge, we introduce Dementia-R1, an RL-based framework for longitudinal dementia prognosis from unstructured clinical notes. Our approach adopts a Cold-Start RL strategy that pre-trains the model to predict verifiable clinical indices extracted from patient histories, enhancing the capability to reason about disease progression before determining the final clinical status. Extensive experiments show that Dementia-R1 achieves the best overall performance on the AMC real-world unstructured cohort, reaching an AUROC of 84.02% and outperforming models up to 10x larger. The framework also generalizes to Parkinson's disease dementia prediction in an independent hospital cohort, achieving an AUROC of 78.37%. On the ADNI benchmark, our 7B model attains the highest AUROC among all LLM baselines at 83.17%, demonstrating strong longitudinal reasoning over fluctuating cognitive trajectories. Code is available at https://anonymous.4open.science/r/dementiar1-CDB5.

cs.CL↗

arXiv2025

FlexiQ: Adaptive Mixed-Precision Quantization for Latency/Accuracy Trade-Offs in Deep Neural Networks

Neural networks commonly execute on hardware accelerators such as NPUs and GPUs for their size and computation overhead. These accelerators are costly and it is hard to scale their resources to handle real-time workload fluctuations. We present FlexiQ, an adaptive mixed-precision quantization scheme for computer vision models. FlexiQ selectively applies low-bitwidth computation to feature channels with small value ranges and employs an efficient bit-lowering method to minimize quantization errors while maintaining inference accuracy. Furthermore, FlexiQ adjusts its low-bitwidth channel ratio in real time, enabling quantized models to effectively manage fluctuating inference workload. We implemented FlexiQ prototype, including the mixed-precision inference runtime on our custom NPU and GPUs. Evaluated on eleven convolution- and transformer-based vision models, FlexiQ achieves on average 6.6% higher accuracy for 4-bit models with finetuning and outperforms four state-of-the-art quantization techniques. Moreover, our mixed-precision models achieved an efficient accuracy-latency trade-off, with the 50% 4-bit model incurring only 0.6% accuracy loss while achieving 40% of the speedup of the 100% 4-bit model over 8-bit model. Latency evaluations on our NPU and GPUs confirmed that FlexiQ introduces minimal runtime overhead, demonstrating its hardware efficiency and overall performance benefits.

cs.LG↗