arXiv subjects

Junho Lee

Publications and source records attributed to Junho Lee.

At least 19 recordsLinked to original sources

arXiv2026

Localizing a Dirac operator via $J$-holomorphic curves

Let $(M,J)$ be a compact almost hermitian $4$-manifold with a smooth embedded $J$-holomorphic curve $C$ representing the canonical class. Motivated by the symplectic Bogomolov--Miyaoka--Yau conjecture, we choose a twisted spin$^{\text{c}}$ Dirac operator ${\mathcal D}$ on $M$ satisfying $$ \operatorname{ind}{\mathcal D}=3c_2(M)-c_1^2(M). $$ We then use the curve $C$ to construct a complex-linear perturbation ${\mathcal A}$ of ${\mathcal D}$ whose singular set is $Z_{\alpha}\sqcup C$, where $Z_{\alpha}=\alpha^{-1}(0)$ for a transverse section $\alpha$ of $\Lambda^{1,0}M$ nonvanishing on $C$. Applying Maridakis' index localization theorem, we express $\operatorname{ind}{\mathcal D}$ as the sum of localized contributions: $3c_2(M)$ from $Z_{\alpha}$ and $-c_1^2(M)$ from $C$.

math.DG↗

arXiv2026

AHEAD: Anticipatory Hand-Driven Teleoperation via Human Intent Prediction

Direct hand-driven teleoperation maps an operator's hand motion to robot end-effector commands at every frame, enabling precise control, but it requires constant monitoring and correction during approach, grasp, and placement, which can be slow and fatiguing. For repetitive pick-and-place tasks, supervisory (goal-based) teleoperation simplifies this process: the operator specifies goals/waypoints, and the robot executes the motion using planning algorithms. Yet, this introduces latency, as the robot must wait for the next command before it can plan and act. "How can we reduce robot reaction time while lowering operator workload?" To tackle this question, we present AHEAD, a real-time VR teleoperation system that anticipates operator intent to enable proactive, hand-driven control. In a digital twin, the operator performs pick-and-place naturally, using hand motion to convey high-level commands rather than a continuous robot trajectory. AHEAD processes a short window of 3D hand and head signals together with scene context through an attention-based classifier to predict the intended grasp object and placement slot. A state machine converts intent predictions into stable robot goals, enabling early motion while remaining stable under noisy predictions and corrective hand movements. AHEAD's intent prediction module achieves Top1 accuracy: 76% for grasp objects and 76% for target slots. Moreover, our user study shows AHEAD reduces robot reaction latency by 0.6 s (object) and 1.4 s (slot) relative to baselines. Participants also reported lower operator load, indicating faster robot responses while maintaining low operator effort in practice.

cs.RO↗

arXiv2026

Domain Generalization via Text-Anchored Information Bottleneck

Visual recognition models often fail when deployed in new environments. Domain Generalization (DG) addresses this by learning representations that remain invariant to environment-specific variations. Recent approaches increasingly rely on large vision-language models, assuming that preserving their expressive visual representations improves robustness. However, we show that such visual expressiveness can instead propagate spurious cues that tie representations to the training environments, hindering invariant learning. We therefore discard visual guidance and instead treat the language embedding space as the primary source of domain invariance, naturally acting as an information bottleneck that preserves core semantics while suppressing domain-specific variations. Extensive experiments across diverse backbones exhibit state-of-the-art performance and further analyze what makes guidance effective for robust generalization. These findings shift the focus of DG from improving representations to designing supervision that enforces invariance.

cs.CV↗

arXiv2026

HEART: Coordination of Heterogeneous Expert Agents for Physically Grounded Robotic Task Planning

Large Language Models (LLMs) can reason over complex instructions but often fail to satisfy the physical and spatial constraints required for robotic task planning. Recent LLM-based planners directly translate text into action sequences, yet they lack structured reasoning about feasibility, reachability, and logical order, resulting in invalid or incomplete plans. We present a heterogeneous multi-LLM framework that decomposes instructions into atomic reasoning tasks and allocates them to role-specialized expert agents under a token budget for real-world computational and communicational constraints. By combining role-oriented reasoning from heterogeneous agents followed by constraint-driven plan synthesis, HEART validates capability, reachability, and constraint conditions before planning and helps produce physically executable plans while maintaining efficiency. Experiments across different household benchmarks show that HEART consistently improves plan success compared to single-LLM and rule-based planners, demonstrating that heterogeneous LLM collaboration enables robust and scalable robotic task planning under resource constraints.

cs.RO↗

arXiv2026

Geometry-Aware Image Flow Matching

Recent advances in generative models highlight the power of geometry-aware modeling in manifold-constrained settings. Yet, for natural images, the field remains confined to Euclidean assumptions, failing to exploit the potential of intrinsic geometric structures within the data. In this work, we investigate the geometry of natural images and observe that semantic information is predominantly encoded in directional components, while norm components can be approximated by the global average. This property holds across both RGB and latent spaces, suggesting that natural images can be effectively modeled on a hypersphere. Building on this finding, we introduce Spherical Optimal Transport Flow Matching (SOT-CFM), which utilizes angular distance, and Spherical Flow Matching (SFM), which constrains dynamics directly on the manifold. Our experiments demonstrate that these geometry-aware methods achieve superior performance against Euclidean baselines. Ultimately, this work provides a novel perspective that bridges the gap between Riemannian manifold-based modeling and natural image generation.

cs.CV↗

arXiv2026

AnyCamVLA: Zero-Shot Camera Adaptation for Viewpoint Robust Vision-Language-Action Models

Despite remarkable progress in Vision-Language-Action models (VLAs) for robot manipulation, these large pre-trained models require fine-tuning to be deployed in specific environments. These fine-tuned models are highly sensitive to camera viewpoint changes that frequently occur in unstructured environments. In this paper, we propose a zero-shot camera adaptation framework without additional demonstration data, policy fine-tuning, or architectural modification. Our key idea is to virtually adjust test-time camera observations to match the training camera configuration in real-time. For that, we use a recent feed-forward novel view synthesis model which outputs high-quality target view images, handling both extrinsic and intrinsic parameters. This plug-and-play approach preserves the pre-trained capabilities of VLAs and applies to any RGB-based policy. Through extensive experiments on the LIBERO benchmark, our method consistently outperforms baselines that use data augmentation for policy fine-tuning or additional 3D-aware features for visual input. We further validate that our approach constantly enhances viewpoint robustness in real-world robotic manipulation scenarios, including settings with varying camera extrinsics, intrinsics, and freely moving handheld cameras. Project Page: https://heo0224.github.io/AnyCamVLA/

cs.RO↗

arXiv2025

Is There a Better Source Distribution than Gaussian? Exploring Source Distributions for Image Flow Matching

Flow matching has emerged as a powerful generative modeling approach with flexible choices of source distribution. While Gaussian distributions are commonly used, the potential for better alternatives in high-dimensional data generation remains largely unexplored. In this paper, we propose a novel 2D simulation that captures high-dimensional geometric properties in an interpretable 2D setting, enabling us to analyze the learning dynamics of flow matching during training. Based on this analysis, we derive several key insights about flow matching behavior: (1) density approximation can paradoxically degrade performance due to mode discrepancy, (2) directional alignment suffers from path entanglement when overly concentrated, (3) Gaussian's omnidirectional coverage ensures robust learning, and (4) norm misalignment incurs substantial learning costs. Building on these insights, we propose a practical framework that combines norm-aligned training with directionally-pruned sampling. This approach maintains the robust omnidirectional supervision essential for stable flow learning, while eliminating initializations in data-sparse regions during inference. Importantly, our pruning strategy can be applied to any flow matching model trained with a Gaussian source, providing immediate performance gains without the need for retraining. Empirical evaluations demonstrate consistent improvements in both generation quality and sampling efficiency. Our findings provide practical insights and guidelines for source distribution design and introduce a readily applicable technique for improving existing flow matching models. Our code is available at https://github.com/kwanseokk/SourceFM.

cs.CV↗

arXiv2025

Optimizing Quantum Data Embeddings for Ligand-Based Virtual Screening

Effective molecular representations are essential for ligand-based virtual screening. We investigate how quantum data embedding strategies can improve this task by developing and evaluating a family of quantum-classical hybrid embedding approaches. These approaches combine classical neural networks with parameterized quantum circuits in different ways to generate expressive molecular representations and are assessed across two benchmark datasets of different sizes: the LIT-PCBA and COVID-19 collections. Across multiple biological targets and class-imbalance settings, several quantum and hybrid embedding variants consistently outperform classical baselines, especially in limited-data regimes. These results highlight the potential of optimized quantum data embeddings as data-efficient tools for ligand-based virtual screening.

quant-ph↗

arXiv2025

HIV-1 protease cleavage sites detection with a Quantum convolutional neural network algorithm

In this study, we propose a quantum convolutional neural network (QCNN)-based framework with the neural quantum embedding (NQE) to predict HIV-1 protease cleavage sites in amino acid sequences from viral and human proteins. To assess the effectiveness and robustness of our framework, we compared the classification performance against classical neural networks under both noiseless and noisy simulations. Among experimental conditions, the QCNN with the angle and amplitude encoding NQE conditions consistently outperformed classical counterparts in both the similar trainable parameter scale and the different number of qubits (the averaged performance of the 4-qubits and 8-qubits QCNN: 0.9146 and 0.8929 / the averaged performance of the classical neural network: 0.6125 and 0.8278). The QCNN with the NQE showed stable performance under the quantum hardware noise, confirming its applicability to biomedical data analysis with the noise intermediate-scale quantum (NISQ) hardware. This study presents the first application of NQE-augmented QCNNs for HIV-1 cleavage site classification, providing new insights into scalable and noise-resilient quantum machine learning for biomedical data.

quant-ph↗

arXiv2025

A Generic and Efficient Python Runtime Verification System and its Large-scale Evaluation

Runtime verification (RV) now scales for testing thousands of open-source Java projects, helping find hundreds of bugs. The popular Python ecosystem could use such benefits. But, today's Python RV systems are limited to a domain or specification logic, or slow. We propose PyMOP, a generic, extensible, and efficient RV system for Python. PyMOP supports five logics, implements five existing monitoring algorithms, ships with 73 API specs of Python and widely-used libraries, supports three instrumentation strategies, and users can easily add more of these. On 290,133 unit tests in 1,463 GitHub projects, we find mainly that (i) the default monitoring algorithm for Java is often not the fastest for Python; (ii) PyMOP is up to 1,168.3x faster than two recent dynamic analysis systems; and (iii) 44 of 121 bugs that PyMOP helped find so far were fixed by developers. PyMOP's generality and efficiency position it well as an excellent platform for the next advances on RV for Python.

cs.SE↗

arXiv2025

CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triples

Deep learning models often learn and exploit spurious correlations in training data, using these non-target features to inform their predictions. Such reliance leads to performance degradation and poor generalization on unseen data. To address these limitations, we introduce a more general form of counterfactual data augmentation, termed counterbias data augmentation, which simultaneously tackles multiple biases (e.g., gender bias, simplicity bias) and enhances out-of-distribution robustness. We present CoBA: CounterBias Augmentation, a unified framework that operates at the semantic triple level: first decomposing text into subject-predicate-object triples, then selectively modifying these triples to disrupt spurious correlations. By reconstructing the text from these adjusted triples, CoBA generates counterbias data that mitigates spurious patterns. Through extensive experiments, we demonstrate that CoBA not only improves downstream task performance, but also effectively reduces biases and strengthens out-of-distribution resilience, offering a versatile and robust solution to the challenges posed by spurious correlations.

cs.CL↗

arXiv2025

Note on concentration via the conjugate-linear Hodge star operator

We construct conjugate-linear perturbations of twisted spinc Dirac operators on compact almost hermitian manifolds of dimension congruent to 2 or 6 modulo 8, employing the conjugate-linear Hodge star operator rescaled by unit complex numbers depending on degree. These perturbations satisfy the concentration principle.

math.DG↗

arXiv2025

Self-Guided Masked Autoencoder

Masked Autoencoder (MAE) is a self-supervised approach for representation learning, widely applicable to a variety of downstream tasks in computer vision. In spite of its success, it is still not fully uncovered what and how MAE exactly learns. In this paper, with an in-depth analysis, we discover that MAE intrinsically learns pattern-based patch-level clustering from surprisingly early stages of pretraining. Upon this understanding, we propose self-guided masked autoencoder, which internally generates informed mask by utilizing its progress in patch clustering, substituting the naive random masking of the vanilla MAE. Our approach significantly boosts its learning process without relying on any external models or supplementary information, keeping the benefit of self-supervised nature of MAE intact. Comprehensive experiments on various downstream tasks verify the effectiveness of the proposed method.

cs.CV↗

arXiv2025

Latent Diffusion Models with Masked AutoEncoders

In spite of the remarkable potential of Latent Diffusion Models (LDMs) in image generation, the desired properties and optimal design of the autoencoders have been underexplored. In this work, we analyze the role of autoencoders in LDMs and identify three key properties: latent smoothness, perceptual compression quality, and reconstruction quality. We demonstrate that existing autoencoders fail to simultaneously satisfy all three properties, and propose Variational Masked AutoEncoders (VMAEs), taking advantage of the hierarchical features maintained by Masked AutoEncoders. We integrate VMAEs into the LDM framework, introducing Latent Diffusion Models with Masked AutoEncoders (LDMAEs). Our code is available at https://github.com/isno0907/ldmae.

cs.CV↗

arXiv2025

Classifying and Tracking International Aid Contribution Towards SDGs

International aid is a critical mechanism for promoting economic growth and well-being in developing nations, supporting progress toward the Sustainable Development Goals (SDGs). However, tracking aid contributions remains challenging due to labor-intensive data management, incomplete records, and the heterogeneous nature of aid data. Recognizing the urgency of this challenge, we partnered with government agencies to develop an AI model that complements manual classification and mitigates human bias in subjective interpretation. By integrating SDG-specific semantics and leveraging prior knowledge from language models, our approach enhances classification accuracy and accommodates the diversity of aid projects. When applied to a comprehensive dataset spanning multiple years, our model can reveal hidden trends in the temporal evolution of international development cooperation. Expert interviews further suggest how these insights can empower policymakers with data-driven decision-making tools, ultimately improving aid effectiveness and supporting progress toward SDGs.

cs.CY↗

arXiv2024

Finding NeMo: Negative-mined Mosaic Augmentation for Referring Image Segmentation

Referring Image Segmentation is a comprehensive task to segment an object referred by a textual query from an image. In nature, the level of difficulty in this task is affected by the existence of similar objects and the complexity of the referring expression. Recent RIS models still show a significant performance gap between easy and hard scenarios. We pose that the bottleneck exists in the data, and propose a simple but powerful data augmentation method, Negative-mined Mosaic Augmentation (NeMo). This method augments a training image into a mosaic with three other negative images carefully curated by a pretrained multimodal alignment model, e.g., CLIP, to make the sample more challenging. We discover that it is critical to properly adjust the difficulty level, neither too ambiguous nor too trivial. The augmented training data encourages the RIS model to recognize subtle differences and relationships between similar visual entities and to concretely understand the whole expression to locate the right target better. Our approach shows consistent improvements on various datasets and models, verified by extensive experiments.

cs.CV↗

arXiv2024

Quantitative Insights into Large Language Model Usage and Trust in Academia: An Empirical Study

Large Language Models (LLMs) are transforming writing, reading, teaching, and knowledge retrieval in many academic fields. However, concerns regarding their misuse and erroneous outputs have led to varying degrees of trust in LLMs within academic communities. In response, various academic organizations have proposed and adopted policies regulating their usage. However, these policies are not based on substantial quantitative evidence because there is no data about use patterns and user opinion. Consequently, there is a pressing need to accurately quantify their usage, user trust in outputs, and concerns about key issues to prioritize in deployment. This study addresses these gaps through a quantitative user study of LLM usage and trust in academic research and education. Specifically, our study surveyed 125 individuals at a private R1 research university regarding their usage of LLMs, their trust in LLM outputs, and key issues to prioritize for robust usage in academia. Our findings reveal: (1) widespread adoption of LLMs, with 75% of respondents actively using them; (2) a significant positive correlation between trust and adoption, as well as between engagement and trust; and (3) that fact-checking is the most critical concern. These findings suggest a need for policies that address pervasive usage, prioritize fact-checking mechanisms, and accurately calibrate user trust levels as they engage with these models. These strategies can help balance innovation with accountability and help integrate LLMs into the academic environment effectively and reliably.

cs.CY↗

arXiv2024

Scalable Frame Sampling for Video Classification: A Semi-Optimal Policy Approach with Reduced Search Space

Given a video with $T$ frames, frame sampling is a task to select $N \ll T$ frames, so as to maximize the performance of a fixed video classifier. Not just brute-force search, but most existing methods suffer from its vast search space of $\binom{T}{N}$, especially when $N$ gets large. To address this challenge, we introduce a novel perspective of reducing the search space from $O(T^N)$ to $O(T)$. Instead of exploring the entire $O(T^N)$ space, our proposed semi-optimal policy selects the top $N$ frames based on the independently estimated value of each frame using per-frame confidence, significantly reducing the computational complexity. We verify that our semi-optimal policy can efficiently approximate the optimal policy, particularly under practical settings. Additionally, through extensive experiments on various datasets and model architectures, we demonstrate that learning our semi-optimal policy ensures stable and high performance regardless of the size of $N$ and $T$.

cs.CV↗