arXiv subjects

Naoki Takada

Publications and source records attributed to Naoki Takada.

10 recordsLinked to original sources

arXiv2026

Understanding Interpretation Difficulty in Harmful Online Communication: Insights from Cybercrime Communities

Harmful online communication often contains slang, coded terms, abbreviations, and community-specific expressions, which make messages difficult to interpret. This paper presents an exploratory study of interpretation difficulty in Discord chats related to cybercrime. We construct reference interpretations of purposefully selected difficult messages, which were reviewed by an expert. We then use them to evaluate human and large language model (LLM) interpretations under different context conditions. The results show that local context alone is often insufficient for humans, while external knowledge and extended conversational context substantially improve human interpretation. For LLMs, local context also improves interpretation, and the larger model performs better. We further conduct a qualitative error analysis and propose a preliminary classification of factors that make harmful chats difficult to interpret. These findings suggest that harmful-content analysis should treat interpretation as an evidence-integration problem, rather than as message-level classification alone.

cs.CL↗

arXiv2026

DD-GEPA: Prompt Optimization for Dialogue Disentanglement Focusing on Task Instruction and Utterance Representation

Multi-party chat often contains interleaved dialogues because multiple participants can discuss different topics at the same time. Dialogue disentanglement addresses this problem by separating an entangled utterance sequence into coherent dialogues. While large language models (LLMs) are promising for this task, they still struggle with dialogue disentanglement and achieve low accuracy. This paper proposes an automatic prompt optimization for LLM based dialogue disentanglement. We decompose the prompt into three components: task instruction, utterance representation, and output instruction, and optimize them using GEPA, an optimization method for compound AI systems. Experiments on benchmark datasets show that the optimized prompts improve dialogue disentanglement accuracy over the original prompts and can surpass hand crafted prompts.

cs.SE↗

arXiv2025

A thermodynamically consistent free-energy lattice Boltzmann model: Incorporating generalized equilibria derived from the color-gradient approach

We extend the chemical-potential-based free-energy lattice Boltzmann (LB) model of Li et al. [Phys. Rev. E 103, 013304 (2021)] by integrating generalized equilibria, originally formulated for the color-gradient LB model using sixth-order Hermite polynomials [Saito et al., Phys Rev E 108, 065305 (2023)], into a thermodynamically consistent framework. Our model is formulated on a three-dimensional D3Q27 lattice with a central-moment collision scheme, simplifying implementation and improving Galilean invariance. Numerical tests, including flat-interface equilibrium, stationary and moving droplets in free space, and wetting on solid surfaces, confirm the model's capability to accurately simulate multiphase phenomena while maintaining strict thermodynamic consistency.

cond-mat.stat-mech↗

arXiv2023

Generalized equilibria for color-gradient lattice Boltzmann model based on higher-order Hermite polynomials: A simplified implementation with central moments

We propose generalized equilibria of a three-dimensional color-gradient lattice Boltzmann model for two-component two-phase flows using higher-order Hermite polynomials. Although the resulting equilibrium distribution function, which includes a sixth-order term on the velocity, is computationally cumbersome, its equilibrium central moments (CMs) are velocity-independent and have a simplified form. Numerical experiments show that our approach, as in Wen et al. [Phys. Rev. E 100, 023301 (2019)] who consider terms up to third order, improves the Galilean invariance compared to that of the conventional approach. Dynamic problems can be solved with high accuracy at a density ratio of 10; however, the accuracy is still limited to a density ratio of $1\,000$. For lower density ratios, the generalized equilibria benefit from the CM-based multiple-relaxation-time model, especially at very high Reynolds numbers, significantly improving the numerical stability.

physics.comp-ph↗

arXiv2017

Deep-learning-based data page classification for holographic memory

We propose a deep-learning-based classification of data pages used in holographic memory. We numerically investigated the classification performance of a conventional multi-layer perceptron (MLP) and a deep neural network, under the condition that reconstructed page data are contaminated by some noise and are randomly laterally shifted. The MLP was found to have a classification accuracy of 91.58%, whereas the deep neural network was able to classify data pages at an accuracy of 99.98%. The accuracy of the deep neural network is two orders of magnitude better than the MLP.

cs.CV↗

arXiv2016

Autoencoder-based holographic image restoration

We propose a holographic image restoration method using an autoencoder, which is an artificial neural network. Because holographic reconstructed images are often contaminated by direct light, conjugate light, and speckle noise, the discrimination of reconstructed images may be difficult. In this paper, we demonstrate the restoration of reconstructed images from holograms that record page data in holographic memory and QR codes by using the proposed method.

cs.CV↗

arXiv2013

Fast computation of computer-generated hologram using Xeon Phi coprocessor

We report fast computation of computer-generated holograms (CGHs) using Xeon Phi coprocessors, which have massively x86-based processors on one chip, recently released by Intel. CGHs can generate arbitrary light wavefronts, and therefore, are promising technology for many applications: for example, three-dimensional displays, diffractive optical elements, and the generation of arbitrary beams. CGHs incur enormous computational cost. In this paper, we describe the implementations of several CGH generating algorithms on the Xeon Phi, and the comparisons in terms of the performance and the ease of programming between the Xeon Phi, a CPU and graphics processing unit (GPU).

physics.comp-ph↗

arXiv2013

Calculation reduction method for color computer-generated hologram using color space conversion

We report a calculation reduction method for color computer-generated holograms (CGHs) using color space conversion. Color CGHs are generally calculated on RGB space. In this paper, we calculate color CGHs in other color spaces: for example, YCbCr color space. In YCbCr color space, a RGB image is converted to the luminance component (Y), blue-difference chroma (Cb) and red-difference chroma (Cr) components. In terms of the human eye, although the negligible difference of the luminance component is well-recognized, the difference of the other components is not. In this method, the luminance component is normal sampled and the chroma components are down-sampled. The down-sampling allows us to accelerate the calculation of the color CGHs. We compute diffraction calculations from the components, and then we convert the diffracted results in YCbCr color space to RGB color space.

physics.optics↗

arXiv2011

Computational wave optics library for C++: CWO++ library

Diffraction calculations, such as the angular spectrum method, and Fresnel diffractions, are used for calculating scalar light propagation. The calculations are used in wide-ranging optics fields: for example, computer generated holograms (CGHs), digital holography, diffractive optical elements, microscopy, image encryption and decryption, three-dimensional analysis for optical devices and so on. However, increasing demands made by large-scale diffraction calculations have rendered the computational power of recent computers insufficient. We have already developed a numerical library for diffraction calculations using a graphic processing unit (GPU), which was named the GWO library. However, this GWO library is not user-friendly, since it is based on C language and was also run only on a GPU. In this paper, we develop a new C++ class library for diffraction and CGH calculations, which is referred as to a CWO++ library, running on a CPU and GPU. We also describe the structure, performance, and usage examples of the CWO++ library.

physics.optics↗

arXiv2010

Fast calculation of computer-generated-hologram on AMD HD5000 series GPU and OpenCL

In this paper, we report fast calculation of a computer-generated-hologram using a new architecture of the HD5000 series GPU (RV870) made by AMD and its new software development environment, OpenCL. Using a RV870 GPU and OpenCL, we can calculate 1,920 * 1,024 resolution of a CGH from a 3D object consisting of 1,024 points in 30 milli-seconds. The calculation speed realizes a speed approximately two times faster than that of a GPU made by NVIDIA.

physics.optics↗