arXiv subjects
Publications and source records attributed to Jiwoong Kim.
This paper proposes a novel method to estimate parameters in a logistic regression model. After obtaining the estimators, their asymptotic properties are rigorously investigated.
Retrieval-Augmented Generation (RAG) has emerged as a powerful approach for grounding Large Language Model (LLM)-based chatbot responses on external knowledge. However, existing RAG studies typically assume well-structured textual sources (e.g. Wikipedia or curated datasets) and perform retrieval and generation at query time, which can limit their applicability in real-world chatbot scenarios. In this paper, we present HybridRAG, a novel and practical RAG framework towards more accurate and faster chatbot responses. First, HybridRAG ingests raw, unstructured PDF documents containing complex layouts (text, tables, figures) via Optical Character Recognition (OCR) and layout analysis, and convert them into hierarchical text chunks. Then, it pre-generates a plausible question-answer (QA) knowledge base from the organized chunks using an LLM. At query time, user questions are matched against this QA bank to retrieve immediate answers when possible, and only if no suitable QA match is found does our framework fall back to an on-the-fly response generation. Experiments on OHRBench demonstrate that our HybridRAG provides higher answer quality and lower latency compared to a standard RAG baseline. We believe that HybridRAG could be a practical solution for real-world chatbot applications that must handle large volumes of unstructured documents and lots of users under limited computational resources.
When the rate parameter of the exponential distribution is associated with predictors, then the main interest will be how to estimate the regression parameter. In this paper, we will investigate how to estimate the parameter on the regression setup of the exponential distribution. To that end, we propose a new estimator, and its asymptotic properties will be discussed.
This paper studies computational aspects of an asymptotically distribution-free goodness-of-fit test for non-Gaussian distributions based on the Khmaladze martingale transformation when the location and scale parameters of the distribution are unknown. On top of that, we propose another goodness-of-fit test better than existing one in terms of a statistical power. Simulation studies demonstrate that the proposed test compares favorably with the existing test.
We report the largest scale deep learning with High Performance Computing (HPC) to physics analysis with the CMS simulation data in proton-proton collisions at 13 TeV. We build a Convolutional Neural Network (CNN) model that takes low-level information as images considering the geometry of the CMS detector and use this model to discriminate \textit{R}-parity violating super symmetry (RPV SUSY) events from the background events with inelastic quantum process from the Standard Model (QCD multi-jet). We compare the classification performance of the CNN method with that of the widely used cut-based method. The signal efficiency (and expected significance) of the CNN method is 1.85 (1.2) times higher than that of the cut-based method. To speed-up the training, the model training is conducted using the Nurion HPC system at the Korea Institute of Science and Technology Information, which is equipped with thousands of parallel \texttt{Xeon Phi} CPUs. Notably, our CNN model shows scalability up to 1024 nodes.
Minimum distance estimation methodology based on an empirical distribution function has been popular due to its desirable properties including robustness. Even though the statistical literature is awash with the research on the minimum distance estimation, the most of it is confined to the theoretical findings: only few statisticians conducted research on the application of the method to real world problems. Through this paper, we extend the domain of application of this methodology to various applied fields by providing a solution to a rather challenging and complicated computational problem. The problem this paper tackles is an image segmentation which has been used in various fields. We propose a novel method based on the classical minimum distance estimation theory to solve the image segmentation problem. The performance of the proposed method is then further elevated by integrating it with the ``segmenting-together" strategy. We demonstrate that the proposed method combined with the segmenting-together strategy successfully completes the segmentation problem when it is applied to the complex, real images such as magnetic resonance images.
The search for topological superconductivity has motivated investigations into materials that combine topological and superconducting properties. The half-Heusler compound YPtBi appears to be such a material, however experiments have thus far been limited to bulk single crystals, drastically limiting the scope of available experiments. This has made it impossible to investigate the potential topological nature of the superconductivity in this material. Experiments to access details about the superconducting state require sophisticated lithographic structures, typically based on thin films. Here we report on the establishment of high crystalline quality epitaxial thin films of YPtBi(111), grown using molecular beam epitaxy on Al2O3(0001) substrates. A robust superconducting state is observed, with both critical temperature and critical field consistent with that previously reported for bulk crystals. Moreover we find that AlOx capping sufficiently protects the sample surface from degradation to allow for proper lithography. Our results pave a path towards the development of advanced lithographic structures, that will allow the exploration of the potentially topological nature of superconductivity in YPtBi.
Chiral symmetry breaking of phonons plays an essential role in emergent quantum phenomena owing to its strong coupling to spin degree of freedom. However, direct experimental evidence of the chiral phonon-spin coupling is lacking. In this study, we report a chiral phonon-mediated interlayer exchange interaction in atomically controlled ferromagnetic metal (SrRuO3)-nonmagnetic insulator (SrTiO3) heterostructures. Owing to the unconventional interlayer exchange interaction, we have observed rotation of magnetic moments as a function of nonmagnetic insulating spacer thickness, resulting in a spin spiral state. The chiral phonon-spin coupling is further confirmed by phonon Zeeman effects. The existence of the chiral phonons and their interplay with spins along with our atomic-scale heterostructure approach open a window to unveil the crucial roles of chiral phonons in magnetic materials.
Artificial crystals synthesized by atomic-scale epitaxy provides the ability to control the dimensions of the quantum phases and associated phase transitions via precise thickness modulation. In particular, reduction in dimensionality via quantized control of atomic layers is a powerful approach to revealing hidden electronic and magnetic phases. Here, we demonstrate a dimensionality-controlled and induced metal-insulator transition (MIT) in atomically designed superlattices by synthesizing a genuine two dimensional (2D) SrRuO3 crystal with highly suppressed charge transfer. The tendency to ferromagnetically align the spins in SrRuO3 layer diminishes in 2D as the interlayer exchange interaction vanishes, accompanying the 2D localization of electrons. Furthermore, electronic and magnetic instabilities in the two SrRuO3 unit cell layers induce a thermally-driven MIT along with a metamagnetic transition.
Ferroelectricity occurs in crystals with broken spatial inversion symmetry. In conventional perovskite oxides, concerted ionic displacements within a three-dimensional network of transition metal-oxygen polyhedra (MOx) manifest spontaneous polarization. Meanwhile, some two-dimensional networks of MOx can foster geometric ferroelectricity with magnetism, owing to the distortion of the polyhedra. Because of the fundamentally different mechanism of ferroelectricity in a two-dimensional network, one can further challenge an uncharted mechanism of ferroelectricity in a one-dimensional channel of MOx and estimate its feasibility. This communication presents ferroelectricity and coupled ferromagnetism in a one-dimensional FeO4 tetrahedral chain network of a brownmillerite SrFeO2.5 epitaxial thin film. The result provides a new paradigm for designing low-dimensional MOx networks, which is expected to benefit the realization of macroscopic ferro-ordering materials including ferroelectric ferromagnets.
The direction of ferroelectric polarization is prescribed by the symmetry of the crystal structure. Therefore, rotation of the polarization direction is largely limited, despite the opportunity it offers in understanding important dielectric phenomena such as piezoelectric response near the morphotropic phase boundaries and practical applications such as ferroelectric memory. In this study, we report the observation of continuous rotation of ferroelectric polarization in order-disorder type LiNbO3 thin films. The spontaneous polarization could be tilted from an out-of-plane to an in-plane direction in the thin film by controlling the Li vacancy concentration within the hexagonal lattice framework. Partial inclusion of monoclinic-like phase is attributed to the breaking of macroscopic inversion symmetry along different directions and the emergence of ferroelectric polarization along the in-plane direction.
This article investigates a fast and stable method to solve Henderson's mixed model equation. The proposed algorithm is stable in that it avoids inverting a matrix of a large dimension and hence is free from the curse of dimensionality. This tactic is enabled through row operations performed on the design matrix.
Elemental defects in transition metal oxides is an important and intriguing subject that result in modifications in variety of physical properties including atomic and electronic structure, optical and magnetic properties. Understanding the formation of elemental vacancies and their influence on different physical properties is essential in studying the complex oxide thin films. In this study, we investigated the physical properties of epitaxial SrRuO3 thin films by systematically manipulating cation and/or oxygen vacancies, via changing the oxygen partial pressure (P(O2)) during the pulsed laser epitaxy (PLE) growth. Ru vacancies in the low-P(O2)-grown SrRuO3 thin films induce lattice expansion with the suppression of the ferromagnetic TC down to ~120 K. Sr vacancies also disturb the ferromagnetic ordering, even though Sr is not a magnetic element. Our results indicate that both A and B cation vacancies in an ABO3 perovskite can be systematically engineered via PLE, and the structural, electrical, and magnetic properties can be tailored accordingly.
Many probability distributions belong to a location-scale family. One such an example is a normal distribution. Probability distributions of the family possess many desirable merits and have colossal amount of real-world application. In this paper, we propose a novel method to estimate both location and scale parameters of probability distributions and investigate its asymptotic properties. After examining theoretical aspects of the proposed estimators, we will empirically demonstrate that they compares favorably with other estimators through simulation studies.
This paper discusses minimum distance estimation method in the linear regression model with dependent errors which are strongly mixing. The regression parameters are estimated through the minimum distance estimation method, and asymptotic distributional properties of the estimators are discussed. A simulation study compares the performance of the minimum distance estimator with other well celebrated estimator. This simulation study shows the superiority of the minimum distance estimator over another estimator. KoulMde (R package) which was used for the simulation study is available online. See section 4 for the detail.
This paper proposes a novel method to estimate the rate parameter of the Poisson distribution. The proposed method employs the Cramer-von Mises type optimization which has been commonly used in estimating parameters of continuous distributions. Upon obtaining the estimator through the proposed method, its desirable properties such as asymptotic distribution and robustness are rigorously investigated. Simulation studies serve to demonstrate that the proposed method compares favorably with other well-celebrated methods including the maximum likelihood method.
The paper proposes a specification test based on two estimates of distribution function. One is the traditional kernel distribution function estimate and the other is a newly proposed convolution-type distribution function estimate. Asymptotic properties of the new estimate are studied when the innovation density is known and when it is unknown. The MISE-type statistic based on these estimates is suggested to test parametric specifications of the mean and volatility functions. The relating asymptotic results are obtained and the finite-sample properties are studied based on the bootstrap methodology. A simulation study shows that the proposed test competes favorably to benchmark tests in terms of the empirical level and power.