arXiv subjects

Hiroshi Ishii

Publications and source records attributed to Hiroshi Ishii.

12 recordsLinked to original sources

arXiv2026

AromaGen: Interactive Generation of Rich Olfactory Experiences with Multimodal Language Models

Smell's connections with food, memory, and social experience have motivated researchers to bring olfaction into interactive systems. Multimodal AI opens new possibilities for generating smells from natural language, yet it remains unclear whether pretrained models encode sufficient olfactory knowledge to translate language into perceptible compositions. We present AromaGen, an AI-powered 12-channel wearable olfactory system that maps free-form natural-language descriptions to 12 base odorants selected to cover a semantically derived olfactory space, while supporting iterative refinement through natural-language feedback. In a between-subjects study (N=60) across 50 real-world benchmark smells, participants distinguished AromaGen-generated target smells above chance in a three-alternative forced-choice task, with retrieval-augmented AI composition performing comparably to human-expert composition and zero-shot AI composition. Natural-language feedback significantly improved the perceived similarity of both AI- and human-expert compositions. Our findings demonstrate the feasibility of using pretrained multimodal AI, with human aroma composition data, to generate perceptible olfactory experiences from natural language.

cs.HC↗

arXiv2026

On the approximation of spatial convolutions by PDE systems

This paper develops an approximation framework for spatial convolutions with radial kernels using elliptic PDE systems in arbitrary spatial dimensions. We prove that radial kernels in Sobolev spaces can be approximated arbitrarily well by linear sums of Green functions associated with a prescribed sequence of positive parameters. The proof is based on a completeness result for an auxiliary system of linear sums of Green functions, established through the identity theorem for analytic functions and a uniqueness argument by integral transforms. The Sobolev approximation provides control of derivatives of the kernels and yields corresponding approximations of convolution operators. As an application, we show that solutions of an aggregation-confinement-diffusion equation can be approximated on any finite time interval by solutions of a corresponding local parabolic-elliptic Keller-Segel system. A numerical example illustrates the kernel approximation.

math.AP↗

arXiv2026

Global front solutions converging to an asymptotic traveling wave in a time-fractional Amari model

We study front propagation in a time-fractional Amari neural field equation with a Caputo derivative and a Heaviside firing-rate function. For a class of nonnegative even connectivity kernels and strictly decreasing $C^1$ front-like initial data, we reduce the Cauchy problem to a nonlinear Volterra equation for the threshold location. We prove the existence and uniqueness of a global front solution, determine the short-time behavior of its threshold location, and identify its asymptotic speed. We also construct the associated asymptotic traveling-wave profile and establish uniqueness of the speed and profile, up to translation, in the nonzero-speed case. Using a rigidity argument, we prove the asymptotic linearity of the local increments of the threshold location and obtain uniform convergence to the profile in the threshold-centered frame. Finally, we characterize the kernel-dependent tail asymptotics of the wave profile.

math.AP↗

arXiv2026

Relationship between haptotaxis and chemotaxis in cell dynamics

Cell sorting mediated by direct cell--cell contact or cellular protrusions can be described by nonlocal cell--cell adhesion models of haptotaxis type, whereas communication through diffusible chemical signals is described by Keller--Segel type chemotaxis systems. We investigate the mathematical relationship between these two descriptions. We first prove that, in the fast signal diffusion limit, subsequences of weak solutions of a Keller--Segel type parabolic--parabolic chemotaxis system with possibly degenerate nonlinear diffusion and density saturation converge to weak solutions of the corresponding parabolic--elliptic system. Eliminating the elliptic chemical fields rewrites the population equation as a nonlocal adhesion model whose interaction kernel is a finite linear combination of Green functions. We then prove, in arbitrary spatial dimensions, that the periodization of the gradient of a prescribed radially symmetric interaction potential can be approximated by finite linear combinations of gradients of Green functions. Combining these results, we prove that suitable weak solutions of the nonlocal adhesion model and a parabolic--parabolic chemotaxis system can be chosen with an arbitrarily small difference over any fixed finite time interval. Numerical simulations compare the nonlocal, parabolic--elliptic, and parabolic--parabolic models and examine how the number of kernel terms and the relaxation time affect their differences.

math.AP↗

arXiv2026

Smell with Genji: Rediscovering Human Perception through an Olfactory Game with AI

Olfaction plays an important role in human perception, yet its subjective and ephemeral nature makes it difficult to articulate, compare, and share across individuals. Traditional practices like the Japanese incense game Genji-ko offer one way to structure olfactory experience through shared interpretation. In this work, we present Smell with Genji, an AI-mediated olfactory interaction system that reinterprets Genji-ko as a collaborative human-AI sensory experience. By integrating a game setup, a mobile application, and an LLM-powered co-smelling partner equipped with olfactory sensing and LLM-based conversation, the system invites participants to compare scents and construct Genji-mon patterns, fostering reflection through a dialogue that highlights the alignment and discrepancies between human and machine perception. This work illustrates how sensing-enabled AI can participate in olfactory experience alongside users, pointing toward new possibilities for AI-supported sensory interaction and reflection in HCI.

cs.HC↗

arXiv2026

Spot solutions to a neural field equation on oblate spheroids

Understanding the dynamics of excitation patterns in neural fields is an important topic in neuroscience. Neural field equations are mathematical models that describe the excitation dynamics of interacting neurons to perform the theoretical analysis. Although many analyses of neural field equations focus on the effect of neuronal interactions on the flat surface, the geometric constraint of the dynamics is also an attractive topic when modeling organs such as the brain. This paper reports pattern dynamics in a neural field equation defined on spheroids as model curved surfaces. We treat spot solutions as localized patterns and discuss how the geometric properties of the curved surface change their properties. To analyze spot patterns on spheroids with small flattening, we first construct exact stationary spot solutions on the spherical surface and reveal their stability. We then extend the analysis to show the existence and stability of stationary spot solutions in the spheroidal case. One of our theoretical results is the derivation of a stability criterion for stationary spot solutions localized at poles on oblate spheroids. The criterion determines whether a spot solution remains at a pole or moves away. Finally, we conduct numerical simulations to discuss the dynamics of spot solutions with the insight of our theoretical predictions. Our results show that the dynamics of spot solutions depend on the curved surface and the coordination of neural interactions.

nlin.PS↗

arXiv2026

Propagating front solutions in a time-fractional Fisher-KPP equation

In this paper, we treat the Fisher-KPP equation with a Caputo-type time fractional derivative and discuss the propagation speed of the solution. The equation is a mathematical model that describes the processes of sub-diffusion, proliferation, and saturation. We first consider a traveling wave solution to study the propagation of the solution, but we cannot define it in the usual sense due to the time fractional derivative in the equation. We therefore assume that the solution asymptotically approaches a traveling wave solution, and the asymptotic traveling wave solution is formally introduced as a potential asymptotic form of the solution. The existence and the properties of the asymptotic traveling wave solution are discussed using a monotone iteration method. Finally, the behavior of the solution is analyzed by numerical simulations based on the result for asymptotic traveling wave solutions.

math.AP↗

arXiv2025

Reaction, diffusion and nonlocal interactions in high-dimensional space

In this paper we consider the mathematical relationship between nonlocal interactions of convolution type and multiple diffusive substances in high dimensions. Motivated by that the nonlocal evolution equations reproduce similar patterns to those in reaction-diffusion systems, we approximate nonlocal interactions in evolution equations by the solution to a reaction-diffusion system in any dimensional Euclidean space. The key aspect of this approach is that any absolutely integrable radial kernels can be approximated by a linear combination of specific Green functions. This enables us to demonstrate that any nonlocal interactions of convolution type can be approximated by a linear sum of auxiliary diffusive substances. Moreover, we show that the parameters in the reaction-diffusion system can be specified depending on the kernel shape up to three dimensions. Our results establish a connection between a broad class of nonlocal interactions and diffusive chemical reactions in dynamical systems.

math.AP↗

arXiv2025

Atlas of Human-AI Interaction (v1): An Interactive Meta-Science Platform for Large-Scale Research Literature Sensemaking

Human-AI interaction researchers face an overwhelming challenge: synthesizing insights from thousands of empirical studies to understand how AI impacts people and inform effective design. Existing approach for literature reviews cluster papers by similarities, keywords or citations, missing the crucial cause-and-effect relationships that reveal how design decisions impact user outcomes. We introduce the Atlas of Human-AI Interaction, an interactive web interface that provides the first systematic mapping of empirical findings across 1,000+ HCI papers using LLM-powered knowledge extraction. Our approach identifies causal relationships, and visualizes them through an AI-enabled interactive web interface as a navigable knowledge graph. We extracted 2,037 empirical findings, revealing research topic clusters, common themes, and disconnected areas. Expert evaluation with 20 researchers revealed the system's effectiveness for discovering research gaps. This work demonstrates how AI can transform literature synthesis itself, offering a scalable framework for evidence-based design, opening new possibilities for computational meta-science across HCI and beyond.

cs.HC↗

arXiv2023

Asymptotic profiles of zero points of solutions to the heat equation

In this paper, we consider the asymptotic profiles of zero points for the spatial variable of the solutions to the heat equation. By giving suitable conditions for the initial data, we prove the existence of zero points by extending the high-order asymptotic expansion theory for the heat equation. This reveals a previously unknown asymptotic profile of zero points diverging at $O(t)$. In a one-dimensional spatial case, we show the zero point's second and third-order asymptotic profiles in a general situation. We also analyze a zero-level set in high-dimensional spaces and obtain results that extend the results for the one-dimensional spatial case.

math.AP↗

arXiv2020

Effective nonlocal kernels on Reaction-diffusion networks

A new method to derive an essential integral kernel from any given reaction-diffusion network is proposed. Any network describing metabolites or signals with arbitrary many factors can be reduced to a single or a simpler system of integro-differential equations called "effective equation" including the reduced integral kernel (called "effective kernel" ) in the convolution type. As one typical example, the Mexican hat shaped kernel is theoretically derived from two component activator-inhibitor systems. It is also shown that a three component system with quite different appearance from activator-inhibitor systems is reduced to an effective equation with the Mexican hat shaped kernel. It means that the two different systems have essentially the same effective equations and that they exhibit essentially the same spatial and temporal patterns. Thus, we can identify two different systems with the understanding in unified concept through the reduced effective kernels. Other two applications of this method are also given: Applications to pigment patterns on skins (two factors network with long range interaction) and waves of differentiation (called proneural waves) in visual systems on brains (four factors network with long range interaction). In the applications, we observe the reproduction of the same spatial and temporal patterns as those appearing in pre-existing models through the numerical simulations of the effective equations.

math.AP↗

arXiv2017

Organic Primitives: Synthesis and Design of pH-Reactive Materials using Molecular I/O for Sensing, Actuation, and Interaction

In this paper we present Organic Primitives, an enabling toolbox that expands upon the library of input-output devices in HCI and facilitates the design of interactions with organic, fluid-based systems. We formulated color, odor and shape changing material primitives which act as sensor-actuators that convert pH signals into human-readable outputs. Food-grade organic molecules anthocyanin, vanillin, and chitosan were employed as dopants to synthesize materials which output a spectrum of colors, degrees of shape deformation, and switch between odorous and non-odorous states. We evaluated the individual output properties of our sensor-actuators to assess the rate, range, and reversibility of the changes as a function of pH 2-10. We present a design space with techniques for enhancing the functionality of the material primitives, and offer passive and computational methods for controlling the material interfaces. Finally, we explore applications enabled by Organic Primitives under four contexts: environmental, cosmetic, edible, and interspecies.

cs.HC↗