arXiv subjects

Junseok Lee

Publications and source records attributed to Junseok Lee.

At least 37 records · Page 2Linked to original sources

arXiv2025

Oldie but Goodie: Re-illuminating Label Propagation on Graphs with Partially Observed Features

In real-world graphs, we often encounter missing feature situations where a few or the majority of node features, e.g., sensitive information, are missed. In such scenarios, directly utilizing Graph Neural Networks (GNNs) would yield sub-optimal results in downstream tasks such as node classification. Despite the emergence of a few GNN-based methods attempting to mitigate its missing situation, when only a few features are available, they rather perform worse than traditional structure-based models. To this end, we propose a novel framework that further illuminates the potential of classical Label Propagation (Oldie), taking advantage of Feature Propagation, especially when only a partial feature is available. Now called by GOODIE, it takes a hybrid approach to obtain embeddings from the Label Propagation branch and Feature Propagation branch. To do so, we first design a GNN-based decoder that enables the Label Propagation branch to output hidden embeddings that align with those of the FP branch. Then, GOODIE automatically captures the significance of structure and feature information thanks to the newly designed Structure-Feature Attention. Followed by a novel Pseudo-Label contrastive learning that differentiates the contribution of each positive pair within pseudo-labels originating from the LP branch, GOODIE outputs the final prediction for the unlabeled nodes. Through extensive experiments, we demonstrate that our proposed model, GOODIE, outperforms the existing state-of-the-art methods not only when only a few features are available but also in abundantly available situations. Source code of GOODIE is available at: https://github.com/SukwonYun/GOODIE.

cs.LG↗

arXiv2025

Global Context-aware Representation Learning for Spatially Resolved Transcriptomics

Spatially Resolved Transcriptomics (SRT) is a cutting-edge technique that captures the spatial context of cells within tissues, enabling the study of complex biological networks. Recent graph-based methods leverage both gene expression and spatial information to identify relevant spatial domains. However, these approaches fall short in obtaining meaningful spot representations, especially for spots near spatial domain boundaries, as they heavily emphasize adjacent spots that have minimal feature differences from an anchor node. To address this, we propose Spotscape, a novel framework that introduces the Similarity Telescope module to capture global relationships between multiple spots. Additionally, we propose a similarity scaling strategy to regulate the distances between intra- and inter-slice spots, facilitating effective multi-slice integration. Extensive experiments demonstrate the superiority of Spotscape in various downstream tasks, including single-slice and multi-slice scenarios. Our code is available at the following link: https: //github.com/yunhak0/Spotscape.

cs.LG↗

arXiv2025

Towards LLM-Centric Multimodal Fusion: A Survey on Integration Strategies and Techniques

The rapid progress of Multimodal Large Language Models(MLLMs) has transformed the AI landscape. These models combine pre-trained LLMs with various modality encoders. This integration requires a systematic understanding of how different modalities connect to the language backbone. Our survey presents an LLM-centric analysis of current approaches. We examine methods for transforming and aligning diverse modal inputs into the language embedding space. This addresses a significant gap in existing literature. We propose a classification framework for MLLMs based on three key dimensions. First, we examine architectural strategies for modality integration. This includes both the specific integration mechanisms and the fusion level. Second, we categorize representation learning techniques as either joint or coordinate representations. Third, we analyze training paradigms, including training strategies and objective functions. By examining 125 MLLMs developed between 2021 and 2025, we identify emerging patterns in the field. Our taxonomy provides researchers with a structured overview of current integration techniques. These insights aim to guide the development of more robust multimodal integration strategies for future models built on pre-trained foundations.

cs.CL↗

arXiv2025

Stability of domain walls with inflationary fluctuations under potential bias, and gravitational wave signatures

A recent study has shown that domain walls with inflationary initial fluctuations exhibit remarkable stability against population bias due to long-range correlations, challenging the claims of prior research. In this paper, we study the dynamics of these domain walls in the presence of potential bias and show that they collapse with a lifetime several times longer than that due to thermal fluctuations. This is interpreted as a difference in the average distance between domain walls, leading us to derive a new formula for the domain wall lifetime which depends on the area parameter in a qualitatively different way from previous studies. In addition, we compute the spectrum of gravitational waves generated by such domain walls and find that both the peak frequency and the peak abundance are lowered in a manner that depends on the area parameter. Based on these findings, we also determine the necessary degree of vacuum degeneracy for axion domain walls to explain the isotropic cosmic birefringence.

hep-ph↗

arXiv2025

Subgraph Federated Learning for Local Generalization

Federated Learning (FL) on graphs enables collaborative model training to enhance performance without compromising the privacy of each client. However, existing methods often overlook the mutable nature of graph data, which frequently introduces new nodes and leads to shifts in label distribution. Since they focus solely on performing well on each client's local data, they are prone to overfitting to their local distributions (i.e., local overfitting), which hinders their ability to generalize to unseen data with diverse label distributions. In contrast, our proposed method, FedLoG, effectively tackles this issue by mitigating local overfitting. Our model generates global synthetic data by condensing the reliable information from each class representation and its structural information across clients. Using these synthetic data as a training set, we alleviate the local overfitting problem by adaptively generalizing the absent knowledge within each local dataset. This enhances the generalization capabilities of local models, enabling them to handle unseen data effectively. Our model outperforms baselines in our proposed experimental settings, which are designed to measure generalization power to unseen data in practical scenarios. Our code is available at https://github.com/sung-won-kim/FedLoG

cs.LG↗

arXiv2025

More is Different: Multi-Axion Dynamics Changes Topological Defect Evolution

We study topological defects in multi-axion models arising from multiple Peccei-Quinn (PQ) scalars. Using a simplified two-axion system, we reveal fundamental differences in the evolution of these defects compared to single-axion scenarios. This finding is particularly significant because, despite the fact that integrating out heavier axions reduces these models to an effective single PQ scalar theory at low energies, the actual physical behavior of topological defects differs markedly from single-axion predictions. Unlike single-axion models where conventional cosmic strings form, multi-axion scenarios with post-inflationary or mixed initial conditions generically produce networks of strings interconnected by high-tension domain walls. This results in a severe cosmological domain wall problem. We determine string-wall network instability conditions and discuss cosmological implications including the application to the QCD axion and gravitational wave generation. Our findings highlight that multi-axion dynamics can lead to qualitatively different outcomes for topological defects, challenging the conventional picture of cosmic evolution of topological defects based on single-axion models.

hep-ph↗

arXiv2025

3rd Workshop on Maritime Computer Vision (MaCVi) 2025: Challenge Results

The 3rd Workshop on Maritime Computer Vision (MaCVi) 2025 addresses maritime computer vision for Unmanned Surface Vehicles (USV) and underwater. This report offers a comprehensive overview of the findings from the challenges. We provide both statistical and qualitative analyses, evaluating trends from over 700 submissions. All datasets, evaluation code, and the leaderboard are available to the public at https://macvi.org/workshop/macvi25.

cs.CV↗

arXiv2024

Induced Domain Walls of QCD Axion, and Gravitational Waves

We show that heavy axion domain walls induce domain walls of the QCD axion through a mixing between the heavy axion and the QCD axion, even when the pre-inflationary initial condition is assumed for the QCD axion. The induced domain walls arise because the effective $θ$ parameter changes across the heavy axion domain walls, shifting the potential minimum of the QCD axion. When the heavy axion domain walls collapse, the induced QCD axion domain walls collapse as well. This novel mechanism for producing the QCD axions can explain dark matter even with the axion decay constant as small as ${\cal O}(10^{9})$ GeV. In particular, this scenario requires domain wall collapse near the QCD crossover, potentially accounting for the stochastic gravitational wave background suggested by recent pulsar timing array observations, including NANOGrav. Using this mechanism, it is also possible to easily create induced domain walls for string axions or axions with a large decay constant, which would otherwise be challenging. We also comment on the implications for cosmic birefringence using induced axion domain walls.

hep-ph↗

arXiv2024

SoccerNet 2024 Challenges Results

The SoccerNet 2024 challenges represent the fourth annual video understanding challenges organized by the SoccerNet team. These challenges aim to advance research across multiple themes in football, including broadcast video understanding, field understanding, and player understanding. This year, the challenges encompass four vision-based tasks. (1) Ball Action Spotting, focusing on precisely localizing when and which soccer actions related to the ball occur, (2) Dense Video Captioning, focusing on describing the broadcast with natural language and anchored timestamps, (3) Multi-View Foul Recognition, a novel task focusing on analyzing multiple viewpoints of a potential foul incident to classify whether a foul occurred and assess its severity, (4) Game State Reconstruction, another novel task focusing on reconstructing the game state from broadcast videos onto a 2D top-view map of the field. Detailed information about the tasks, challenges, and leaderboards can be found at https://www.soccer-net.org, with baselines and development kits available at https://github.com/SoccerNet.

cs.CV↗

arXiv2024

MART: MultiscAle Relational Transformer Networks for Multi-agent Trajectory Prediction

Multi-agent trajectory prediction is crucial to autonomous driving and understanding the surrounding environment. Learning-based approaches for multi-agent trajectory prediction, such as primarily relying on graph neural networks, graph transformers, and hypergraph neural networks, have demonstrated outstanding performance on real-world datasets in recent years. However, the hypergraph transformer-based method for trajectory prediction is yet to be explored. Therefore, we present a MultiscAle Relational Transformer (MART) network for multi-agent trajectory prediction. MART is a hypergraph transformer architecture to consider individual and group behaviors in transformer machinery. The core module of MART is the encoder, which comprises a Pair-wise Relational Transformer (PRT) and a Hyper Relational Transformer (HRT). The encoder extends the capabilities of a relational transformer by introducing HRT, which integrates hyperedge features into the transformer mechanism, promoting attention weights to focus on group-wise relations. In addition, we propose an Adaptive Group Estimator (AGE) designed to infer complex group relations in real-world environments. Extensive experiments on three real-world datasets (NBA, SDD, and ETH-UCY) demonstrate that our method achieves state-of-the-art performance, enhancing ADE/FDE by 3.9%/11.8% on the NBA dataset. Code is available at https://github.com/gist-ailab/MART.

cs.LG↗

arXiv2024

Bubble Misalignment Mechanism for Axions

We study the dynamics of axions at first-order phase transitions in non-Abelian gauge theories. When the duration of the phase transition is short compared to the timescale of the axion oscillations, the axion dynamics is similar to the trapped misalignment mechanism. On the other hand, if this is not the case, the axions are initially expelled from the inside of the bubbles, generating axion waves on the outside. Analogous to the Fermi acceleration, these axions gain energy by repeatedly scattering off the bubble walls. Once they acquire enough energy, they can enter the bubbles. The novel ``bubble misalignment mechanism'' can significantly enhance the axion abundance, compared to models where the axion mass is either constant or varies continuously as a function of temperature. The increase in axion abundance depends on the axion mass, the duration of the phase transition, and the bubble wall velocity. This mechanism results in a spatially inhomogeneous distribution of axions, which could lead to the formation of axion miniclusters. It has potential implications for the formation of oscillons/I-balls, axion warm dark matter, cosmic birefringence, and the production of dark photons.

hep-ph↗

arXiv2023

Gravitational Waves from Domain Wall Collapse, and Application to Nanohertz Signals with QCD-coupled Axions

We study for the first time the gravitational waves generated during the collapse of domain walls, incorporating the potential bias in the lattice simulations. The final stages of domain wall collapse are crucial for the production of gravitational waves, but have remained unexplored due to computational difficulties. As a significant application of this new result, we show that the observed NANOGrav, EPTA, PPTA, and CPTA data, which indicate stochastic gravitational waves in the nanohertz regime, can be attributed to axion domain walls coupled to QCD. In our model, non-perturbative effects of QCD induce a temperature-dependent bias around the QCD crossover, inducing the rapid collapse of the domain walls. We use sophisticated lattice simulations that account for the temperature-dependent bias to measure the gravitational waves resulting from the domain wall annihilation. We also discuss the future prospects for accelerator-based searches for the axion and the potential for the formation and detection of primordial black holes.

hep-ph↗

arXiv2023

M2Former: Multi-Scale Patch Selection for Fine-Grained Visual Recognition

Recently, vision Transformers (ViTs) have been actively applied to fine-grained visual recognition (FGVR). ViT can effectively model the interdependencies between patch-divided object regions through an inherent self-attention mechanism. In addition, patch selection is used with ViT to remove redundant patch information and highlight the most discriminative object patches. However, existing ViT-based FGVR models are limited to single-scale processing, and their fixed receptive fields hinder representational richness and exacerbate vulnerability to scale variability. Therefore, we propose multi-scale patch selection (MSPS) to improve the multi-scale capabilities of existing ViT-based models. Specifically, MSPS selects salient patches of different scales at different stages of a multi-scale vision Transformer (MS-ViT). In addition, we introduce class token transfer (CTT) and multi-scale cross-attention (MSCA) to model cross-scale interactions between selected multi-scale patches and fully reflect them in model decisions. Compared to previous single-scale patch selection (SSPS), our proposed MSPS encourages richer object representations based on feature hierarchy and consistently improves performance from small-sized to large-sized objects. As a result, we propose M2Former, which outperforms CNN-/ViT-based models on several widely used FGVR benchmarks.

cs.CV↗

arXiv2023

Single-cell RNA-seq data imputation using Feature Propagation

While single-cell RNA sequencing provides an understanding of the transcriptome of individual cells, its high sparsity, often termed dropout, hampers the capture of significant cell-cell relationships. Here, we propose scFP (single-cell Feature Propagation), which directly propagates features, i.e., gene expression, especially in raw feature space, via cell-cell graph. Specifically, it first obtains a warmed-up cell-gene matrix via Hard Feature Propagation which fully utilizes known gene transcripts. Then, we refine the k-Nearest Neighbor(kNN) of the cell-cell graph with a warmed-up cell-gene matrix, followed by Soft Feature Propagation which now allows known gene transcripts to be further denoised through their neighbors. Through extensive experiments on imputation with cell clustering tasks, we demonstrate our proposed model, scFP, outperforms various recent imputation and clustering methods. The source code of scFP can be found at https://github.com/Junseok0207/scFP.

cs.CE↗

arXiv2023

Conditional Graph Information Bottleneck for Molecular Relational Learning

Molecular relational learning, whose goal is to learn the interaction behavior between molecular pairs, got a surge of interest in molecular sciences due to its wide range of applications. Recently, graph neural networks have recently shown great success in molecular relational learning by modeling a molecule as a graph structure, and considering atom-level interactions between two molecules. Despite their success, existing molecular relational learning methods tend to overlook the nature of chemistry, i.e., a chemical compound is composed of multiple substructures such as functional groups that cause distinctive chemical reactions. In this work, we propose a novel relational learning framework, called CGIB, that predicts the interaction behavior between a pair of graphs by detecting core subgraphs therein. The main idea is, given a pair of graphs, to find a subgraph from a graph that contains the minimal sufficient information regarding the task at hand conditioned on the paired graph based on the principle of conditional graph information bottleneck. We argue that our proposed method mimics the nature of chemical reactions, i.e., the core substructure of a molecule varies depending on which other molecule it interacts with. Extensive experiments on various tasks with real-world datasets demonstrate the superiority of CGIB over state-of-the-art baselines. Our code is available at https://github.com/Namkyeong/CGIB.

q-bio.MN↗

arXiv2023

Task-Equivariant Graph Few-shot Learning

Although Graph Neural Networks (GNNs) have been successful in node classification tasks, their performance heavily relies on the availability of a sufficient number of labeled nodes per class. In real-world situations, not all classes have many labeled nodes and there may be instances where the model needs to classify new classes, making manual labeling difficult. To solve this problem, it is important for GNNs to be able to classify nodes with a limited number of labeled nodes, known as few-shot node classification. Previous episodic meta-learning based methods have demonstrated success in few-shot node classification, but our findings suggest that optimal performance can only be achieved with a substantial amount of diverse training meta-tasks. To address this challenge of meta-learning based few-shot learning (FSL), we propose a new approach, the Task-Equivariant Graph few-shot learning (TEG) framework. Our TEG framework enables the model to learn transferable task-adaptation strategies using a limited number of training meta-tasks, allowing it to acquire meta-knowledge for a wide range of meta-tasks. By incorporating equivariant neural networks, TEG can utilize their strong generalization abilities to learn highly adaptable task-specific strategies. As a result, TEG achieves state-of-the-art performance with limited training meta-tasks. Our experiments on various benchmark datasets demonstrate TEG's superiority in terms of accuracy and generalization ability, even when using minimal meta-training data, highlighting the effectiveness of our proposed approach in addressing the challenges of meta-learning based few-shot node classification. Our code is available at the following link: https://github.com/sung-won-kim/TEG

cs.LG↗

arXiv2023

Heterogeneous Graph Learning for Multi-modal Medical Data Analysis

Routine clinical visits of a patient produce not only image data, but also non-image data containing clinical information regarding the patient, i.e., medical data is multi-modal in nature. Such heterogeneous modalities offer different and complementary perspectives on the same patient, resulting in more accurate clinical decisions when they are properly combined. However, despite its significance, how to effectively fuse the multi-modal medical data into a unified framework has received relatively little attention. In this paper, we propose an effective graph-based framework called HetMed (Heterogeneous Graph Learning for Multi-modal Medical Data Analysis) for fusing the multi-modal medical data. Specifically, we construct a multiplex network that incorporates multiple types of non-image features of patients to capture the complex relationship between patients in a systematic way, which leads to more accurate clinical decisions. Extensive experiments on various real-world datasets demonstrate the superiority and practicality of HetMed. The source code for HetMed is available at https://github.com/Sein-Kim/Multimodal-Medical.

cs.CV↗

arXiv2022

Teaching Where to Look: Attention Similarity Knowledge Distillation for Low Resolution Face Recognition

Deep learning has achieved outstanding performance for face recognition benchmarks, but performance reduces significantly for low resolution (LR) images. We propose an attention similarity knowledge distillation approach, which transfers attention maps obtained from a high resolution (HR) network as a teacher into an LR network as a student to boost LR recognition performance. Inspired by humans being able to approximate an object's region from an LR image based on prior knowledge obtained from HR images, we designed the knowledge distillation loss using the cosine similarity to make the student network's attention resemble the teacher network's attention. Experiments on various LR face related benchmarks confirmed the proposed method generally improved recognition performances on LR settings, outperforming state-of-the-art results by simply transferring well-constructed attention maps. The code and pretrained models are publicly available in the https://github.com/gist-ailab/teaching-where-to-look.

cs.CV↗