Source author record

Tsuyoshi Kato

Tsuyoshi Kato appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

math.DS math.MG Computer Vision math.DG math.GR math.SG Machine Learning math-ph math.AP math.FA math.GT math.KT math.MP math.SP Neural and Evolutionary Computing nlin.SI

Catalog footprint

What is connected

13works

16topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2022arXiv

Seiberg-Witten theory on finite covering spaces of spin 4-manifolds

We compute the equivariant Bauer-Furuta degree, when a finite group acts freely on a spin 4-manifold. In the case when the group is cyclic of order power of two, Bryan gave a formula and its applications. We have treated the case when the group has order of odd degree. In particular we gave a formula of the degree when the order is odd-prime. Our approach is to use a representation-theoretic method on finite dimensional approximations of the functional spaces.

preprint2020arXiv

$L^2$ harmonic forms and the Seiberg-Witten map on non compact four manifolds

We explain a new phenomenon on non compact complete Riemannian four manifolds, where d^+ image of one forms can not exhaust densely on L^2 self dual forms on each compact subset, if a certain L^2 self dual harmonic form exists. This leads to construct a new functional analytic framework on the Seiberg-Witten map.

preprint2020arXiv

Adaptive Signal Variances: CNN Initialization Through Modern Architectures

Deep convolutional neural networks (CNN) have achieved the unwavering confidence in its performance on image processing tasks. The CNN architecture constitutes a variety of different types of layers including the convolution layer and the max-pooling layer. CNN practitioners widely understand the fact that the stability of learning depends on how to initialize the model parameters in each layer. Nowadays, no one doubts that the de facto standard scheme for initialization is the so-called Kaiming initialization that has been developed by He et al. The Kaiming scheme was derived from a much simpler model than the currently used CNN structure having evolved since the emergence of the Kaiming scheme. The Kaiming model consists only of the convolution and fully connected layers, ignoring the max-pooling layer and the global average pooling layer. In this study, we derived the initialization scheme again not from the simplified Kaiming model, but precisely from the modern CNN architectures, and empirically investigated how the new initialization method performs compared to the de facto standard ones that are widely used today.

preprint2016arXiv

Spectral analysis of transition operators, Automata groups and translation in BBS

We give the automata which describe time evolution rules of the box-ball system (BBS) with a carrier. It can be shown by use of tropical geometry, such systems are ultradiscrete analogues of KdV equation. We discuss their relation with the lamplighter group generated by an automaton. We present spectral analysis of the stochastic matrices induced by these automata, and verify their spectral coincidence.

preprint2016arXiv

Stochastic Dykstra Algorithms for Metric Learning on Positive Semi-Definite Cone

Recently, covariance descriptors have received much attention as powerful representations of set of points. In this research, we present a new metric learning algorithm for covariance descriptors based on the Dykstra algorithm, in which the current solution is projected onto a half-space at each iteration, and runs at O(n^3) time. We empirically demonstrate that randomizing the order of half-spaces in our Dykstra-based algorithm significantly accelerates the convergence to the optimal solution. Furthermore, we show that our approach yields promising experimental results on pattern recognition tasks.

preprint2015arXiv

New Descriptor for Glomerulus Detection in Kidney Microscopy Image

Glomerulus detection is a key step in histopathological evaluation of microscopy images of kidneys. However, the task of automatic detection of glomeruli poses challenges due to the disparity in sizes and shapes of glomeruli in renal sections. Moreover, extensive variations of their intensities due to heterogeneity in immunohistochemistry staining are also encountered. Despite being widely recognized as a powerful descriptor for general object detection, the rectangular histogram of oriented gradients (Rectangular HOG) suffers from many false positives due to the aforementioned difficulties in the context of glomerulus detection. A new descriptor referred to as Segmental HOG is developed to perform a comprehensive detection of hundreds of glomeruli in images of whole kidney sections. The new descriptor possesses flexible blocks that can be adaptively fitted to input images to acquire robustness to deformations of glomeruli. Moreover, the novel segmentation technique employed herewith generates high quality segmentation outputs and the algorithm is assured to converge to an optimal solution. Consequently, experiments using real world image data reveal that Segmental HOG achieves significant improvements in detection performance compared to Rectangular HOG. The proposed descriptor and method for glomeruli detection present promising results and is expected to be useful in pathological evaluation.

preprint2014arXiv

A note on the pentagram map and tropical geometry

In this note we study the action of the pentagram map on the moduli space of twisted polygons. The action with respect to the canonical coordinate turns out to be applicable to the framework of tropical geometry. As an application, we induce quasi-recursiveness of the pentagram map on some domain, which is induced from dynamical property of the pentagram automaton.

preprint2013arXiv

Localized ASD moduli spaces based on the reduced cohomology group over Casson handles

In this paper, we introduce a deformation analysis of index theory over non compact manifolds, by use of new functional spaces which are the reduced version of Sobolev spaces. It allows to construct Fredholm theory for elliptic differential operators over non compact spaces which possibly do not have closed range with respect to the standard Sobolev spaces. We apply it to construct moduli theory mainly over Casson handles, and obtain another approach to see high complexity of their smooth structure.

preprint2012arXiv

Infinite sequences of almost Kaehler manifolds with high symmetry, their perturbations and pseudo holomorphic curves

We study analysis over infinite dimensional manifolds consisted by sequences of almost Kaehler manifolds. We develop moduli theory of pseudo holomorphic curves into such spaces with high symmetry. Many mechanisms of the standard moduli theory over finite dimensional spaces also work over these infinite dimensional spaces, which is based on a simple functional analytic framework.

preprint2011arXiv

Automata in groups and dynamics and induced systems of PDE in tropical geometry

In this paper we develop a dynamical scaling limit from rational dynamics to automata in tropical geometry. We compare these dynamics and induce uniform estimates of their orbits. We apply these estimates to introduce a comparison analysis of theory of automata groups in geometric group theory with analysis of rational dynamics and some PDE systems. Frameworks of characteristic properties of automata groups are inherited to the corresponding rational or PDE dynamics. As an application we study the Burnside problem in group theory and translate the property as the infinite quasi-recursiveness in rational dynamics.

preprint2011arXiv

The Mesoscopic category, Automata and Tropical Geometry

We start with comparisons of hierarchies in Biology and relate it to Quan- tum Field Theories. Thereby we discover many similarities and translate them into rich mathematical correspondences. The basic connection goes via scale transformations and Tropical geometry. One of our core observations is that Cellular Automata can be naturally introduced in many physical models and refer to a (generalised) mesoscopic scale, as we point it out for the fundamental relation with Gauge Field Theories. To illustrate our framework further, we apply it to the Witten-Kontsevich model and the Miller-Morita-Mumford classes, which in our picture arise from a dynamical system.

preprint2010arXiv

A rough analytic relation on partial differential equations

We introduce some analytic relations on the set of partial differential equations of two variables. It relies on a new comparison method to give rough asymptotic estimates for solutions which obey different partial differential equations. It uses a kind of scale transform called tropical geometry which connects automata with real rational dynamics. Two different solutions can be considered when their defining equations are transformed to the same automata at infinity. We have a systematic way to construct related pairs of different partial differential equations, and also construct some unrelated pairs concretely. These verify that the new relations are non trivial. We also make numerical calculations and compare the results for both related and unrelated pairs of PDEs.

preprint2004arXiv

ASD moduli spaces over four-manifolds with tree-like ends

In this paper we construct Riemannian metrics and weight functions over Casson handles. We show that the corresponding Atiyah-Hitchin-Singer complexes are Fredholm for some class of Casson handles of bounded type. Using these, the Yang-Mills moduli spaces are constructed as finite dimensional smooth manifolds over Casson handles in the class.