arXiv subjects

Jaewoo Park

Publications and source records attributed to Jaewoo Park.

At least 37 records · Page 2Linked to original sources

arXiv2026

K-EXAONE Technical Report

This technical report presents K-EXAONE, a large-scale multilingual language model developed by LG AI Research. K-EXAONE is built on a Mixture-of-Experts architecture with 236B total parameters, activating 23B parameters during inference. It supports a 256K-token context window and covers six languages: Korean, English, Spanish, German, Japanese, and Vietnamese. We evaluate K-EXAONE on a comprehensive benchmark suite spanning reasoning, agentic, general, Korean, and multilingual abilities. Across these evaluations, K-EXAONE demonstrates performance comparable to open-weight models of similar size. K-EXAONE, designed to advance AI for a better life, is positioned as a powerful proprietary AI foundation model for a wide range of industrial and research applications.

cs.CL↗

arXiv2025

A Delayed Acceptance Auxiliary Variable MCMC for Spatial Models with Intractable Likelihood Function

A large class of spatial models contains intractable normalizing functions, such as spatial lattice models, interaction spatial point processes, and social network models. Bayesian inference for such models is challenging since the resulting posterior distribution is doubly intractable. Although auxiliary variable MCMC (AVM) algorithms are known to be the most practical, they are computationally expensive due to the repeated auxiliary variable simulations. To address this, we propose delayed-acceptance AVM (DA-AVM) methods, which can reduce the number of auxiliary variable simulations. The first stage of the kernel uses a cheap surrogate to decide whether to accept or reject the proposed parameter value. The second stage guarantees detailed balance with respect to the posterior. The auxiliary variable simulation is performed only on the parameters accepted in the first stage. We construct various surrogates specifically tailored for doubly intractable problems, including subsampling strategy, Gaussian process emulation, and frequentist estimator-based approximation. We validate our method through simulated and real data applications, demonstrating its practicality for complex spatial models.

stat.ME↗

arXiv2025

Explain with Visual Keypoints Like a Real Mentor! A Benchmark for Multimodal Solution Explanation

With the rapid advancement of mathematical reasoning capabilities in Large Language Models (LLMs), AI systems are increasingly being adopted in educational settings to support students' comprehension of problem-solving processes. However, a critical component remains underexplored in current LLM-generated explanations: multimodal explanation. In real-world instructional contexts, human tutors routinely employ visual aids, such as diagrams, markings, and highlights, to enhance conceptual clarity. To bridge this gap, we introduce the multimodal solution explanation task, designed to evaluate whether models can identify visual keypoints, such as auxiliary lines, points, angles, and generate explanations that incorporate these key elements essential for understanding. To evaluate model performance on this task, we propose ME2, a multimodal benchmark consisting of 1,000 math problems annotated with visual keypoints and corresponding explanatory text that references those elements. Our empirical results show that current models struggle to identify visual keypoints. In the task of generating keypoint-based explanations, open-source models also face notable difficulties. This highlights a significant gap in current LLMs' ability to perform mathematical visual grounding, engage in visually grounded reasoning, and provide explanations in educational contexts. We expect that the multimodal solution explanation task and the ME2 dataset will catalyze further research on LLMs in education and promote their use as effective, explanation-oriented AI tutors.

cs.CL↗

arXiv2025

Snell meets Fagnano. Path optimization through an imperfect mirror

The renowned Fagnano problem asks for the inscribed triangle of minimal perimeter within a given reference triangle. Equivalently, it seeks a billiard trajectory inside the triangle that closes after three reflections. In this note, we consider a modification of this classical problem: finding the inscribed triangle of minimal weighted perimeter -- or, equivalently, the periodic trajectory of a Snell billiard.

math.DS↗

arXiv2025

EP-HDC: Hyperdimensional Computing with Encrypted Parameters for High-Throughput Privacy-Preserving Inference

While homomorphic encryption (HE) provides strong privacy protection, its high computational cost has restricted its application to simple tasks. Recently, hyperdimensional computing (HDC) applied to HE has shown promising performance for privacy-preserving machine learning (PPML). However, when applied to more realistic scenarios such as batch inference, the HDC-based HE has still very high compute time as well as high encryption and data transmission overheads. To address this problem, we propose HDC with encrypted parameters (EP-HDC), which is a novel PPML approach featuring client-side HE, i.e., inference is performed on a client using a homomorphically encrypted model. Our EP-HDC can effectively mitigate the encryption and data transmission overhead, as well as providing high scalability with many clients while providing strong protection for user data and model parameters. In addition to application examples for our client-side PPML, we also present design space exploration involving quantization, architecture, and HE-related parameters. Our experimental results using the BFV scheme and the Face/Emotion datasets demonstrate that our method can improve throughput and latency of batch inference by orders of magnitude over previous PPML methods (36.52~1068x and 6.45~733x, respectively) with less than 1% accuracy degradation.

cs.CR↗

arXiv2025

Posterior Contraction for Sparse Neural Networks in Besov Spaces with Intrinsic Dimensionality

This work establishes that sparse Bayesian neural networks achieve optimal posterior contraction rates over anisotropic Besov spaces and their hierarchical compositions. These structures reflect the intrinsic dimensionality of the underlying function, thereby mitigating the curse of dimensionality. Our analysis shows that Bayesian neural networks equipped with either sparse or continuous shrinkage priors attain the optimal rates which are dependent on the intrinsic dimension of the true structures. Moreover, we show that these priors enable rate adaptation, allowing the posterior to contract at the optimal rate even when the smoothness level of the true function is unknown. The proposed framework accommodates a broad class of functions, including additive and multiplicative Besov functions as special cases. These results advance the theoretical foundations of Bayesian neural networks and provide rigorous justification for their practical effectiveness in high-dimensional, structured estimation problems.

stat.ML↗

arXiv2025

SAIL: SRAM-Accelerated LLM Inference System with Lookup-Table-based GEMV

Large Language Model (LLM) inference requires substantial computational resources, yet CPU-based inference remains essential for democratizing AI due to the widespread availability of CPUs compared to specialized accelerators. However, efficient LLM inference on CPUs faces two fundamental challenges: (1) existing CPU architectures struggle with low-precision arithmetic required by quantized models, where optimal bit precision varies across models and layers; and (2) the memory-bound nature of the token generation phase creates severe performance bottlenecks. To address these challenges, we propose SAIL (SRAM-Accelerated Inference of LLMs), a CPU-based inference solution that efficiently supports arbitrary bit precisions with minimal overhead. SAIL integrates three key innovations: First, we introduce Batched LUT-based General Matrix-Vector Multiplication (LUT-GEMV) with SRAM-based processing-in-memory, enabling high data reuse through lookup tables and reducing memory movement. Second, our Pattern-Aware LUT optimization identifies and exploits redundancy in input activation patterns, reducing computation cycles by 13.8\%. Third, we develop an in-memory type conversion algorithm that leverages PIM's parallelism for efficient de-/quantization operations, alleviating pressure on CPU's vector units. Our architecture requires only 2\% hardware overhead and a single new instruction, while maintaining dual functionality as both compute and storage units. Experimental evaluations using a modified gem5 simulator demonstrate that SAIL achieves up to 10.7x speedup and 19.9x higher tokens per dollar compared to ARM Neoverse-N1 CPU baselines, and up to 7.04x better cost efficiency than NVIDIA V100 GPUs, establishing a practical path for efficient CPU-based LLM inference.

cs.AR↗

arXiv2025

Challenges and Lessons from MIDOG 2025: A Two-Stage Approach to Domain-Robust Mitotic Figure Detection

Mitotic figure detection remains a challenging task in computational pathology due to domain variability and morphological complexity. This paper describes our participation in the MIDOG 2025 challenge, focusing on robust mitotic figure detection across diverse tissue domains. We developed a two-stage pipeline combining Faster R-CNN for candidate detection with an ensemble of three classifiers (DenseNet-121, EfficientNet-v2, InceptionResNet-v2) for false positive reduction. Our best submission achieved F1-score 0.2237 (Recall: 0.9528, Precision: 0.1267) using a Faster R-CNN trained solely on MIDOG++ dataset. While our high recall demonstrates effective mitotic figure detection, the critically low precision (12.67%) reveals fundamental challenges in distinguishing true mitoses from morphologically similar imposters across diverse domains. Analysis of six submission variants showed that subsequent optimization attempts were counterproductive, highlighting the omplexity of domain generalization in histopathology. This work provides valuable insights into the practical challenges of developing robust mitotic figure detection algorithms and emphasizes the importance of effective false positive suppression strategies.

eess.IV↗

arXiv2025

Zero-shot Multimodal Document Retrieval via Cross-modal Question Generation

Rapid advances in Multimodal Large Language Models (MLLMs) have expanded information retrieval beyond purely textual inputs, enabling retrieval from complex real world documents that combine text and visuals. However, most documents are private either owned by individuals or confined within corporate silos and current retrievers struggle when faced with unseen domains or languages. To address this gap, we introduce PREMIR, a simple yet effective framework that leverages the broad knowledge of an MLLM to generate cross modal pre questions (preQs) before retrieval. Unlike earlier multimodal retrievers that compare embeddings in a single vector space, PREMIR leverages preQs from multiple complementary modalities to expand the scope of matching to the token level. Experiments show that PREMIR achieves state of the art performance on out of distribution benchmarks, including closed domain and multilingual settings, outperforming strong baselines across all retrieval metrics. We confirm the contribution of each component through in depth ablation studies, and qualitative analyses of the generated preQs further highlight the model's robustness in real world settings.

cs.IR↗

arXiv2025

A Stein Gradient Descent Approach for Doubly Intractable Distributions

Bayesian inference for doubly intractable distributions is challenging because they include intractable terms, which are functions of parameters of interest. Although several alternatives have been developed for such models, they are computationally intensive due to repeated auxiliary variable simulations. We propose a novel Monte Carlo Stein variational gradient descent (MC-SVGD) approach for inference for doubly intractable distributions. Through an efficient gradient approximation, our MC-SVGD approach rapidly transforms an arbitrary reference distribution to approximate the posterior distribution of interest, without necessitating any predefined variational distribution class for the posterior. Such a transport map is obtained by minimizing Kullback-Leibler divergence between the transformed and posterior distributions in a reproducing kernel Hilbert space (RKHS). We also investigate the convergence rate of the proposed method. We illustrate the application of the method to challenging examples, including a Potts model, an exponential random graph model, and a Conway--Maxwell--Poisson regression model. The proposed method achieves substantial computational gains over existing algorithms, while providing comparable inferential performance for the posterior distributions.

stat.ML↗

arXiv2025

RefPose: Leveraging Reference Geometric Correspondences for Accurate 6D Pose Estimation of Unseen Objects

Estimating the 6D pose of unseen objects from monocular RGB images remains a challenging problem, especially due to the lack of prior object-specific knowledge. To tackle this issue, we propose RefPose, an innovative approach to object pose estimation that leverages a reference image and geometric correspondence as guidance. RefPose first predicts an initial pose by using object templates to render the reference image and establish the geometric correspondence needed for the refinement stage. During the refinement stage, RefPose estimates the geometric correspondence of the query based on the generated references and iteratively refines the pose through a render-and-compare approach. To enhance this estimation, we introduce a correlation volume-guided attention mechanism that effectively captures correlations between the query and reference images. Unlike traditional methods that depend on pre-defined object models, RefPose dynamically adapts to new object shapes by leveraging a reference image and geometric correspondence. This results in robust performance across previously unseen objects. Extensive evaluation on the BOP benchmark datasets shows that RefPose achieves state-of-the-art results while maintaining a competitive runtime.

cs.CV↗

arXiv2025

Bayesian Function-on-Function Regression for Spatial Functional Data

Spatial functional data arise in many settings, such as particulate matter curves observed at monitoring stations and age population curves at each areal unit. Most existing functional regression models have limited applicability because they do not consider spatial correlations. Although functional kriging methods can predict the curves at unobserved spatial locations, they are based on variogram fittings rather than constructing hierarchical statistical models. In this manuscript, we propose a Bayesian framework for spatial function-on-function regression that can carry out parameter estimations and predictions. However, the proposed model has computational and inferential challenges because the model needs to account for within and between-curve dependencies. Furthermore, high-dimensional and spatially correlated parameters can lead to the slow mixing of Markov chain Monte Carlo algorithms. To address these issues, we first utilize a basis transformation approach to simplify the covariance and apply projection methods for dimension reduction. We also develop a simultaneous band score for the proposed model to detect the significant region in the regression function. We apply our method to both areal and point-level spatial functional data, showing the proposed method is computationally efficient and provides accurate estimations and predictions.

stat.ME↗

arXiv2025

Background-Aware Defect Generation for Robust Industrial Anomaly Detection

Detecting anomalies in industrial settings is challenging due to the scarcity of labeled anomalous data. Generative models can mitigate this issue by synthesizing realistic defect samples, but existing approaches often fail to model the crucial interplay between defects and their background. This oversight leads to unrealistic anomalies, especially in scenarios where contextual consistency is essential (i.e., logical anomaly). To address this, we propose a novel background-aware defect generation framework, where the background influences defect denoising without affecting the background itself by ensuring realistic synthesis while preserving structural integrity. Our method leverages a disentanglement loss to separate the background' s denoising process from the defect, enabling controlled defect synthesis through DDIM Inversion. We theoretically demonstrate that our approach maintains background fidelity while generating contextually accurate defects. Extensive experiments on MVTec AD and MVTec Loco benchmarks validate our mehtod's superiority over existing techniques in both defect generation quality and anomaly detection performance.

cs.CV↗

arXiv2024

Face Reconstruction Transfer Attack as Out-of-Distribution Generalization

Understanding the vulnerability of face recognition systems to malicious attacks is of critical importance. Previous works have focused on reconstructing face images that can penetrate a targeted verification system. Even in the white-box scenario, however, naively reconstructed images misrepresent the identity information, hence the attacks are easily neutralized once the face system is updated or changed. In this paper, we aim to reconstruct face images which are capable of transferring face attacks on unseen encoders. We term this problem as Face Reconstruction Transfer Attack (FRTA) and show that it can be formulated as an out-of-distribution (OOD) generalization problem. Inspired by its OOD nature, we propose to solve FRTA by Averaged Latent Search and Unsupervised Validation with pseudo target (ALSUV). To strengthen the reconstruction attack on OOD unseen encoders, ALSUV reconstructs the face by searching the latent of amortized generator StyleGAN2 through multiple latent optimization, latent optimization trajectory averaging, and unsupervised validation with a pseudo target. We demonstrate the efficacy and generalization of our method on widely used face datasets, accompanying it with extensive ablation studies and visually, qualitatively, and quantitatively analyses. The source code will be released.

cs.CV↗

arXiv2024

A More Accurate Approximation of Activation Function with Few Spikes Neurons

Recent deep neural networks (DNNs), such as diffusion models [1], have faced high computational demands. Thus, spiking neural networks (SNNs) have attracted lots of attention as energy-efficient neural networks. However, conventional spiking neurons, such as leaky integrate-and-fire neurons, cannot accurately represent complex non-linear activation functions, such as Swish [2]. To approximate activation functions with spiking neurons, few spikes (FS) neurons were proposed [3], but the approximation performance was limited due to the lack of training methods considering the neurons. Thus, we propose tendency-based parameter initialization (TBPI) to enhance the approximation of activation function with FS neurons, exploiting temporal dependencies initializing the training parameters.

cs.NE↗

arXiv2024

Finding Most Shattering Minimum Vertex Cuts of Polylogarithmic Size in Near-Linear Time

We show the first near-linear time randomized algorithms for listing all minimum vertex cuts of polylogarithmic size that separate the graph into at least three connected components (also known as shredders) and for finding the most shattering one, i.e., the one maximizing the number of connected components. Our algorithms break the quadratic time bound by Cheriyan and Thurimella (STOC'96) for both problems that has stood for more than two decades. Our work also removes a bottleneck to near-linear time algorithms for the vertex connectivity augmentation problem (Jordan '95). Note that it is necessary to list only minimum vertex cuts that separate the graph into at least three components because there can be an exponential number of minimum vertex cuts in general. To obtain near-linear time algorithms, we have extended techniques in local flow algorithms developed by Forster et al. (SODA'20) to list shredders on a local scale. We also exploit fast queries to a pairwise vertex connectivity oracle subject to vertex failures (Long and Saranurak FOCS'22, Kosinas ESA'23). This is the first application of connectivity oracles subject to vertex failures to speed up a static graph algorithm.

cs.DS↗

arXiv2024

Impacts of Innovation School System in Korea: A Latent Space Item Response Model with Neyman-Scott Point Process

South Korea's educational system has faced criticism for its lack of focus on critical thinking and creativity, resulting in high levels of stress and anxiety among students. As part of the government's effort to improve the educational system, the innovation school system was introduced in 2009, which aims to develop students' creativity as well as their non-cognitive skills. To better understand the differences between innovation and regular school systems in South Korea, we propose a novel method that combines the latent space item response model (LSIRM) with the Neyman-Scott (NS) point process model. Our method accounts for the heterogeneity of items and students, captures relationships between respondents and items, and identifies item and student clusters that can provide a comprehensive understanding of students' behaviors/perceptions on non-cognitive outcomes. Our analysis reveals that students in the innovation school system show a higher sense of citizenship, while those in the regular school system tend to associate confidence in appearance with social ability. We compare our model with exploratory item factor analysis in terms of item clustering and find that our approach provides a more detailed and automated analysis. A comparison with exploratory item factor analysis highlights our method's advantages in terms of uncertainty quantification of the clustering process and more detailed and nuanced clustering results. Our method is made available to an existing R package, lsirm12pl.

stat.AP↗

arXiv2024

A Bayesian Convolutional Neural Network-based Generalized Linear Model

Convolutional neural networks (CNNs) provide flexible function approximations for a wide variety of applications when the input variables are in the form of images or spatial data. Although CNNs often outperform traditional statistical models in prediction accuracy, statistical inference, such as estimating the effects of covariates and quantifying the prediction uncertainty, is not trivial due to the highly complicated model structure and overparameterization. To address this challenge, we propose a new Bayesian approach by embedding CNNs within the generalized linear models (GLMs) framework. We use extracted nodes from the last hidden layer of CNN with Monte Carlo (MC) dropout as informative covariates in GLM. This improves accuracy in prediction and regression coefficient inference, allowing for the interpretation of coefficients and uncertainty quantification. By fitting ensemble GLMs across multiple realizations from MC dropout, we can account for uncertainties in extracting the features. We apply our methods to biological and epidemiological problems, which have both high-dimensional correlated inputs and vector covariates. Specifically, we consider malaria incidence data, brain tumor image data, and fMRI data. By extracting information from correlated inputs, the proposed method can provide an interpretable Bayesian analysis. The algorithm can be broadly applicable to image regressions or correlated data analysis by enabling accurate Bayesian inference quickly.

stat.ME↗