A5013378061- Fractional Differential Equations Solutions
- Complex Systems and Time Series Analysis
- Stochastic processes and financial applications
- Cosmology and Gravitation Theories
- Quantum Electrodynamics and Casimir Effect
- Advanced Thermodynamics and Statistical Mechanics
- Quantum Mechanics and Applications
- Financial Risk and Volatility Modeling
- stochastic dynamics and bifurcation
- Statistical Mechanics and Entropy
- Nonlinear Differential Equations Analysis
- Chaos control and synchronization
- Black Holes and Theoretical Physics
- Machine Learning and Algorithms
- Quantum Chromodynamics and Particle Interactions
- Complex Network Analysis Techniques
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Nonlinear Partial Differential Equations
- Advanced Bandit Algorithms Research
- Noncommutative and Quantum Gravity Theories
- Machine Learning and Data Classification
- Reinforcement Learning in Robotics
- Particle physics theoretical and experimental studies
- Gaussian Processes and Bayesian Inference
Korea University
2015-2025
Statistics Korea
2023-2025
Pukyong National University
2023
Ulsan National Institute of Science and Technology
2018-2022
Kao Corporation (Japan)
2020
Brain (Germany)
2020
Korea Kacoh (South Korea)
2019
Korea University
2019
Multimedia University
2007-2017
Chulalongkorn University
2011
We study some Gaussian models for anomalous diffusion, which include the time-rescaled Brownian motion, two types of fractional and associated with motion based on generalized Langevin equation. processes these satisfy diffusion relation requires mean-square displacement to vary t(alpha), 0<alpha<2. However, have different properties, thus indicating that a single parameter is insufficient characterize underlying mechanism. Although versions all same probability distribution function,...
In this paper, we propose an uncertainty-aware learning from demonstration method by presenting a novel uncertainty estimation utilizing mixture density network appropriate for modeling complex and noisy human behaviors. The proposed acquisition can be done with single forward path without Monte Carlo sampling is suitable real-time robotics applications. Then, show that it decomposed into explained variance unexplained where the connections between aleatoric epistemic uncertainties are...
Distributed-order fractional Langevin-like equations are introduced and applied to describe anomalous diffusion without unique or scaling exponent. It is shown that these Langevin of distributed order can be used model the kinetics retarding subdiffusion whose exponent decreases with time strongly ultraslow mean square displacement which varies asymptotically as a power logarithm time.
Performance of data-driven network for tumor classification varies with stain-style histopathological images. This article proposes the transfer (SST) model based on conditional generative adversarial networks (GANs) which is to learn not only certain color distribution but also corresponding pattern. Our considers feature-preserving loss in addition well-known GAN loss. Consequently our does transfers initial stain-styles desired one prevent degradation classifier transferred The examined...
Oscillator of single‐degree‐freedom is a typical model in system analysis. Oscillations resulted from differential equations with fractional order attract the interests researchers since such type oscillations may appear dramatic behaviors responses. However, solution to impulse response class oscillators studied this paper remains unknown field. In paper, we propose closed form oscillators. Based on it, reveal stability behavior as follows. A oscillator be strictly stable, nonstable, or...
We introduce a new fractional oscillator process which can be obtained as solution of stochastic differential equation with two orders. Basic properties such fractal dimension and short-range dependence the are studied by considering asymptotic its covariance function. By velocity diffusion process, we derive corresponding constant, fluctuation–dissipation relation mean-square displacement. The also regarded one-dimensional Euclidean Klein–Gordon field, applying Parisi–Wu quantization method...
We derive rigorously explicit formulas of the Casimir free energy at finite temperature for massless scalar field and electromagnetic confined in a closed rectangular cavity with different boundary conditions by zeta regularization method.We study both low high expansions energy.In each case, we write as sum polynomial plus exponentially decay terms.We show that is always decreasing function temperature.In cases Dirichlet condition field, zero might be positive.In these cases, there unique...
Let p(t, x) be the fundamental solution to problem <TEX>$${\partial}^{\alpha}_tu=-(-{\Delta})^{\beta}u,\;{\alpha}{\in}(0,2),\;{\beta}{\in}(0,{\infty})$$</TEX>. If <TEX>${\alpha},{\beta}{\in}(0,1)$</TEX>, then kernel becomes transition density of a Levy process delayed by an inverse subordinator. In this paper we provide asymptotic behaviors and sharp upper bounds its space time fractional derivatives...
Many important applications, including robotics, data-center management, and process control, require planning action sequences in domains with continuous state spaces discontinuous objective functions. Monte Carlo tree search (MCTS) is an effective strategy for discrete spaces. We provide a novel MCTS algorithm (voot) deterministic environments spaces, which, turn, based on black-box function-optimization (voo) to efficiently sample actions. The voo uses Voronoi partitioning guide sampling,...
We design and implement a ready-to-use library in PyTorch for performing micro-batch pipeline parallelism with checkpointing proposed by GPipe (Huang et al., 2019). In particular, we develop set of components to enable pipeline-parallel gradient computation PyTorch's define-by-run eager execution environment. show that each component is necessary fully benefit from such environment, demonstrate the efficiency applying it various network architectures including AmoebaNet-D U-Net. Our...
We study some of the basic properties a generalized Cauchy process indexed by two parameters. The application Lamperti transformation to leads self-similar which preserves long-range dependence. asymptotic spectral density are derived. Possible this model relaxation phenomena is considered.
We developed a machine learning algorithm (MLA) that can classify human thyroid cell clusters by exploiting both Papanicolaou staining and intrinsic refractive index (RI) as correlative imaging contrasts evaluated the effects of this combination on diagnostic performance. Thyroid fine-needle aspiration biopsy (FNAB) specimens were analyzed using optical diffraction tomography, which simultaneously measure both, color brightfield three-dimensional RI distribution. The MLA was designed to...
Recent advancements in AI have revolutionized property prediction materials science and accelerating material discovery. Graph neural networks (GNNs) stand out due to their ability represent crystal structures as graphs, effectively capturing local interactions delivering superior predictions. However, these methods often lose critical global information, such systems repetitive unit connectivity. To address this, we propose CAST, a cross-attention-based multimodal fusion model that...