arXiv subjects

Kyungeun Kim

Publications and source records attributed to Kyungeun Kim.

15 recordsLinked to original sources

arXiv2026

RaLMPH: Reliability-aware Learning for Multi-Pathologist Harmonization in Whole-Slide Image Classification

Multiple Instance Learning (MIL) is a standard paradigm for Whole-Slide Image (WSI) analysis and has achieved strong results in computational pathology. However, most MIL pipelines assume a single "gold" label per slide, which conflicts with clinical practice where substantial inter-pathologist variability is common. Existing multi-annotator learning and label-refinement methods typically estimate global annotator reliability or rely on single-instance assumptions, making them poorly suited to MIL and to localized diagnostic contexts where experts disagree. We propose RaLMPH (Reliability-aware Learning for Multi-Pathologist Harmonization), a MIL-based label reconciliation framework for WSIs annotated by multiple pathologists. RaLMPH introduces a reliability field that jointly models (i) local neighborhood structure in WSI feature space and (ii) expert uncertainty (entropy), enabling per-sample identification of trustworthy reference neighborhoods. Leveraging this field, RaLMPH performs sample-wise local annotator ranking to select reliable opinions per slide and applies an adaptive gating mechanism to fuse labels conditioned on local reliability. Experiments on a clinical WSI dataset with labels from six pathologists, as well as controlled simulated benchmarks, show that RaLMPH consistently outperforms existing approaches. Further analyses clarify how our reliability-aware mechanism improves label reconciliation and downstream MIL performance.

cs.CV↗

arXiv2026

Perturbative Contrastive Physical Learning

Responses to perturbations are key to understanding physical systems. The ability to contrast such responses by comparing how a system reacts under slightly different conditions provides a mechanism for learning. Here, we introduce Perturbative Contrastive Physical Learning (PCPL), a general framework in which learning emerges from measurable contrasts between physical states produced by controlled changes to inputs, boundary conditions, parameters, or interpreter functions. PCPL unifies and extends prior approaches: Equilibrium Propagation is rooted in contrasts between free and nudged equilibria in energy-based systems, while Frequency Propagation corresponds to contrasts extracted from sinusoidally driven, frequency-demodulated responses. We show that contrast-driven updates can reflect either local sensitivities or global inverse-problem structure, yet do not require centralized gradient computation. Instead, effective learning geometry emerges implicitly from the system's own physical response, allowing learning behavior to arise without an external processor or explicit backpropagation. We demonstrate PCPL in two platforms: (i) spring networks that update bond stiffness using measured displacements and forces, and (ii) continuous-variable photonic circuits trained via x quadrature measurements and finite-difference estimates of the Jacobian. Both platforms successfully learn classification tasks. We further show that a continuous-variable photonic circuit can be trained to implement analog multiplication, illustrating a step toward more autonomous physical learning systems.

cs.LG↗

arXiv2026

Every Error has Its Magnitude: Asymmetric Mistake Severity Training for Multiclass Multiple Instance Learning

Multiple Instance Learning (MIL) has emerged as a promising paradigm for Whole Slide Image (WSI) diagnosis, offering effective learning with limited annotations. However, existing MIL frameworks overlook diagnostic priorities and fail to differentiate the severity of misclassifications in multiclass, leaving clinically critical errors unaddressed. We propose a mistake-severity-aware training strategy that organizes diagnostic classes into a hierarchical structure, with each level optimized using a severity-weighted cross-entropy loss that penalizes high-severity misclassifications more strongly. Additionally, hierarchical consistency is enforced through probabilistic alignment, a semantic feature remix applied to the instance bag to robustly train class priority and accommodate clinical cases involving multiple symptoms. An asymmetric Mikel's Wheel-based metric is also introduced to quantify the severity of errors specific to medical fields. Experiments on challenging public and real-world in-house datasets demonstrate that our approach significantly mitigates critical errors in MIL diagnosis compared to existing methods. We present additional experimental results on natural domain data to demonstrate the generalizability of our proposed method beyond medical contexts.

cs.CV↗

arXiv2026

Laser interferometry as a robust neuromorphic platform for machine learning

We present a method for implementing an optical neural network using only linear optical resources, namely field displacement and interferometry applied to coherent states of light. The nonlinearity required for learning in a neural network is realized via an encoding of the input into phase shifts allowing for far more straightforward experimental implementation compared to previous proposals for, and demonstrations of, $\textit{in situ}$ inference. Beyond $\textit{in situ}$ inference, the method enables $\textit{in situ}$ training by utilizing established techniques like parameter shift methods or physical backpropagation to extract gradients directly from measurements of the linear optical circuit. We also investigate the effect of photon losses and find the model to be very resilient to these.

physics.optics↗

arXiv2026

Cell strain-stiffening drives cell breakout from embedded spheroids

Understanding how cells escape from embedded spheroids requires a mechanical framework linking stress generation within cells, across cells, and between cells and the surrounding extracellular matrix (ECM). We develop such a framework by coupling a 3D vertex model of a spheroid to a fibrous ECM network and deriving a 3D Cauchy stress tensor for deformable polyhedral cells, enabling direct cell-level stress quantification in three dimensions. We analyze maximum shear stress in solid-like and fluid-like spheroids: solid-like spheroids exhibit broader stress distributions and radial stress gradients, while fluid-like spheroids show lower stresses with weak spatial organization. Cell shape anisotropy is not generically aligned with principal stress directions, indicating that morphology alone is an unreliable proxy for mechanical state. We further demonstrate strain stiffening at the single-cell level, where elongation produces nonlinear increases in maximum shear stress, allowing boundary cells in otherwise low-stress, fluid-like spheroids to transiently generate forces sufficient to remodel the matrix. To connect strain-induced stress amplification to invasion modes, we introduce an extended 3D vertex model with explicit, tunable cell-cell adhesion springs. In this minimal mechanical framework, single-cell breakout results from strain stiffening combined with reduced adhesion, whereas multi-cell streaming additionally requires anisotropic adhesion strengthened along the elongation axis and weakened orthogonally. Together, these results identify distinct mechanical pathways coupling cell strain, stress amplification, and adhesion organization to spheroid invasion.

physics.bio-ph↗

arXiv2025

Priority-Aware Clinical Pathology Hierarchy Training for Multiple Instance Learning

Multiple Instance Learning (MIL) is increasingly being used as a support tool within clinical settings for pathological diagnosis decisions, achieving high performance and removing the annotation burden. However, existing approaches for clinical MIL tasks have not adequately addressed the priority issues that exist in relation to pathological symptoms and diagnostic classes, causing MIL models to ignore priority among classes. To overcome this clinical limitation of MIL, we propose a new method that addresses priority issues using two hierarchies: vertical inter-hierarchy and horizontal intra-hierarchy. The proposed method aligns MIL predictions across each hierarchical level and employs an implicit feature re-usability during training to facilitate clinically more serious classes within the same level. Experiments with real-world patient data show that the proposed method effectively reduces misdiagnosis and prioritizes more important symptoms in multiclass scenarios. Further analysis verifies the efficacy of the proposed components and qualitatively confirms the MIL predictions against challenging cases with multiple symptoms.

cs.CV↗

arXiv2025

Design Principles for Realizable Discrete Surface Embeddings in Physical Systems

The isometric embedding of surfaces in three-dimensional space is fundamental to various physical systems, from elastic sheets to programmable materials. While continuous surfaces typically admit unique solutions under suitable boundary conditions, their discrete counterparts-represented as networks of vertices connected by edges-can exhibit multiple distinct embeddings for identical edge lengths. We present a systematic approach to constructing discrete meshes that yield a controlled number of embeddings. By analyzing the relationship between mesh connectivity and embedding multiplicity through rigidity theory, we develop criteria for designing meshes that minimize solution multiplicity. We demonstrate computational methods based on local matrix operations and trilateration techniques, enabling practical implementation for meshes with approximately a thousand vertices. Our analysis provides both theoretical bounds on the number of possible embeddings based on Bézout's theorem and practical guidelines for mesh construction in physical applications. Through numerical simulations, we show that this approach achieves comparable accuracy to traditional minimization methods while offering computational advantages through sequential computation. Importantly, we demonstrate that in cases where a unique smooth solution exists, local fluctuations in reconstructed shapes derived from the computational grid can serve as indicators of insufficient geometric constraints. This work bridges the gap between discrete and continuous embedding problems, providing insights for applications in 4D printing, mechanical meta-materials, and deployable structures.

cond-mat.dis-nn↗

arXiv2025

VLEER: Vision and Language Embeddings for Explainable Whole Slide Image Representation

Recent advances in vision-language models (VLMs) have shown remarkable potential in bridging visual and textual modalities. In computational pathology, domain-specific VLMs, which are pre-trained on extensive histopathology image-text datasets, have succeeded in various downstream tasks. However, existing research has primarily focused on the pre-training process and direct applications of VLMs on the patch level, leaving their great potential for whole slide image (WSI) applications unexplored. In this study, we hypothesize that pre-trained VLMs inherently capture informative and interpretable WSI representations through quantitative feature extraction. To validate this hypothesis, we introduce Vision and Language Embeddings for Explainable WSI Representation (VLEER), a novel method designed to leverage VLMs for WSI representation. We systematically evaluate VLEER on three pathological WSI datasets, proving its better performance in WSI analysis compared to conventional vision features. More importantly, VLEER offers the unique advantage of interpretability, enabling direct human-readable insights into the results by leveraging the textual modality for detailed pathology annotations, providing clear reasoning for WSI-level pathology downstream tasks.

cs.CV↗

arXiv2024

Rigidity condition for gluing two bar-joint rigid graphs embedded in $\mathbb{R}^d$

How does one determine if a collection of bars joined by freely rotating hinges cannot be deformed without changing the length of any of the bars? In other words, how does one determine if a bar-joint graph is rigid? This question has been definitively answered using combinatorial rigidity theory in two dimensions via the Geiringer-Laman Theorem. However, it has not yet been answered using combinatorial rigidity theory in higher dimensions, given known counterexamples to the trivial dimensional extension of the Geiringer-Laman Theorem. To work towards a combinatorial approach in dimensions beyond two, we present a theorem for gluing two rigid bar-joint graphs together that remain rigid. When there are no overlapping vertices between the two graphs, the theorem reduces to Tay's theorem used to identify rigidity in body-bar graphs. When there are overlapping vertices, we rely on the notion of pinned rigid graphs to identify and constrain rigid motions. This theorem provides a basis for an algorithm for recursively constructing rigid clusters that can be readily adapted for computational purposes. By leveraging Henneberg-type operations to grow a rigid (or minimally rigid) graph and treating simplices-where every vertex connects to every other vertex-as fundamental units, our approach offers a scalable solution with computational complexity comparable to traditional methods. Thus, we provide a combinatorial blueprint for algorithms in multi-dimensional rigidity theory as applied to bar-joint graphs.

cond-mat.dis-nn↗

arXiv2024

CAMP: Continuous and Adaptive Learning Model in Pathology

There exist numerous diagnostic tasks in pathology. Conventional computational pathology formulates and tackles them as independent and individual image classification problems, thereby resulting in computational inefficiency and high costs. To address the challenges, we propose a generic, unified, and universal framework, called a continuous and adaptive learning model in pathology (CAMP), for pathology image classification. CAMP is a generative, efficient, and adaptive classification model that can continuously adapt to any classification task by leveraging pathology-specific prior knowledge and learning taskspecific knowledge with minimal computational cost and without forgetting the knowledge from the existing tasks. We evaluated CAMP on 22 datasets, including 1,171,526 patches and 11,811 pathology slides, across 17 classification tasks. CAMP achieves state-of-theart classification performance on a wide range of datasets and tasks at both patch- and slide-levels and reduces up to 94% of computation time and 85% of storage memory in comparison to the conventional classification models. Our results demonstrate that CAMP can offer a fundamental transformation in pathology image classification, paving the way for the fully digitized and computerized pathology practice.

eess.IV↗

arXiv2024

DIOR-ViT: Differential Ordinal Learning Vision Transformer for Cancer Classification in Pathology Images

In computational pathology, cancer grading has been mainly studied as a categorical classification problem, which does not utilize the ordering nature of cancer grades such as the higher the grade is, the worse the cancer is. To incorporate the ordering relationship among cancer grades, we introduce a differential ordinal learning problem in which we define and learn the degree of difference in the categorical class labels between pairs of samples by using their differences in the feature space. To this end, we propose a transformer-based neural network that simultaneously conducts both categorical classification and differential ordinal classification for cancer grading. We also propose a tailored loss function for differential ordinal learning. Evaluating the proposed method on three different types of cancer datasets, we demonstrate that the adoption of differential ordinal learning can improve the accuracy and reliability of cancer grading, outperforming conventional cancer grading approaches. The proposed approach should be applicable to other diseases and problems as they involve ordinal relationship among class labels.

eess.IV↗

arXiv2024

A two-dimensional vertex model for curvy cell-cell interfaces at the subcellular scale

Cross-sections of cell shapes in a tissue monolayer typically resemble a tiling of convex polygons. Yet, examples exist where the polygons are not convex with curved cell-cell interfaces, as seen in the adaxial epidermis. To date, two-dimensional vertex models predicting the structure and mechanics of cell monolayers have been mostly limited to convex polygons. To overcome this limitation, we introduce a framework to study curvy cell-cell interfaces at the subcellular scale within vertex models by using a parameterized curve between vertices that is expanded in a Fourier series and whose coefficients represent additional degrees of freedom. This extension to non-convex polygons allows for cells with same shape index, or dimensionless perimeter, to be, for example, either elongated or globular with lobes. In the presence of applied, anisotropic stresses, we find that local, subcellular curvature, or buckling, can be energetically more favorable than larger scale deformations involving groups of cells. Inspired by recent experiments, we also find that local, subcellular curvature at cell-cell interfaces emerges in a group of cells in response to the swelling of additional cells surrounding the group. Our framework, therefore, can account for a wider array of multi-cellular responses to constraints in the tissue environment.

physics.bio-ph↗

arXiv2024

Enhanced extracellular matrix remodeling due to embedded spheroid fluidization

Tumor spheroids are in vitro three-dimensional, cellular collectives consisting of cancerous cells. Embedding these spheroids in an in vitro fibrous environment, such as a collagen network, to mimic the extracellular matrix (ECM) provides an essential platform to quantitatively investigate the biophysical mechanisms leading to tumor invasion of the ECM. To understand the mechanical interplay between tumor spheroids and the ECM, we computationally construct and study a three-dimensional vertex model for a tumor spheroid that is mechanically coupled to a cross-linked network of fibers. In such a vertex model, cells are represented as deformable polyhedrons that share faces. Some fraction of the boundary faces of the tumor spheroid contain linker springs connecting the center of the boundary face to the nearest node in the fiber network. As these linker springs actively contract, the fiber network remodels. By toggling between fluid-like and solid-like spheroids via changing the dimensionless cell shape index, we find that the spheroid rheology affects the remodeling of the fiber network. More precisely, fluid-like spheroids displace the fiber network more on average near the vicinity of the spheroid than solid-like spheroids. We also find more densification of the fiber network near the spheroid for the fluid-like spheroids. These spheroid rheology-dependent effects are the result of cellular motility due to active cellular rearrangements that emerge over time in the fluid-like spheroids to generate spheroid shape fluctuations. Our results uncover intricate morphological-mechanical interplay between an embedded spheroid and its surrounding fiber network with both spheroid contractile strength and spheroid shape fluctuations playing important roles in the pre-invasion stages of tumor invasion.

physics.bio-ph↗

arXiv2023

Mean field elastic moduli of a three-dimensional cell-based vertex model

The mechanics of a foam typically depends on the bubble geometry, topology, and the material at hand, be it metallic or polymeric, for example. While the foam energy functional for each bubble is typically minimization of surface area for a given volume, biology provides us with a wealth of additional energy functionals, should one consider biological cells as a foam-like material. Here, we focus on a mean field approach to obtain the elastic moduli, within linear response, for an ordered, three-dimensional vertex model using the space-filling shape of a truncated octahedron and whose energy functional is characterized by a restoring surface area spring and a restoring volume spring. The tuning of the three-dimensional shape index exhibits a rigidity transition via a compatible-incompatible transition. Specifically, for smaller shape indices, both the target surface area and volume cannot be achieved, while beyond some critical value of the three-dimensional shape index, they can be, resulting in a zero-energy state. As the elastic moduli depend on curvatures of the energy when the system, we obtain these as well. In addition to analytically determining the location of the transition in mean field, we find that the rigidity transition and the elastic moduli depend on the parameterization of the cell shape with this effect being more pronounced in three dimensions given the array of shapes that a polyhedron can take on (as compared to a polygon). We also uncover nontrivial dependence on the deformation protocol in which some deformations result in affine motion of the vertices, while others result in nonaffine motion. Such dependencies on the shape parameterization and deformation protocol give rise to a nontrivial shape landscape and, therefore, nontrivial mechanical response even in the absence of topology changes.

cond-mat.soft↗

arXiv2022

GradMix for nuclei segmentation and classification in imbalanced pathology image datasets

An automated segmentation and classification of nuclei is an essential task in digital pathology. The current deep learning-based approaches require a vast amount of annotated datasets by pathologists. However, the existing datasets are imbalanced among different types of nuclei in general, leading to a substantial performance degradation. In this paper, we propose a simple but effective data augmentation technique, termed GradMix, that is specifically designed for nuclei segmentation and classification. GradMix takes a pair of a major-class nucleus and a rare-class nucleus, creates a customized mixing mask, and combines them using the mask to generate a new rare-class nucleus. As it combines two nuclei, GradMix considers both nuclei and the neighboring environment by using the customized mixing mask. This allows us to generate realistic rare-class nuclei with varying environments. We employed two datasets to evaluate the effectiveness of GradMix. The experimental results suggest that GradMix is able to improve the performance of nuclei segmentation and classification in imbalanced pathology image datasets.

eess.IV↗