arXiv subjects

Takashi Sato

Publications and source records attributed to Takashi Sato.

At least 37 records · Page 2Linked to original sources

arXiv2023

Modular DFR: Digital Delayed Feedback Reservoir Model for Enhancing Design Flexibility

A delayed feedback reservoir (DFR) is a type of reservoir computing system well-suited for hardware implementations owing to its simple structure. Most existing DFR implementations use analog circuits that require both digital-to-analog and analog-to-digital converters for interfacing. However, digital DFRs emulate analog nonlinear components in the digital domain, resulting in a lack of design flexibility and higher power consumption. In this paper, we propose a novel modular DFR model that is suitable for fully digital implementations. The proposed model reduces the number of hyperparameters and allows flexibility in the selection of the nonlinear function, which improves the accuracy while reducing the power consumption. We further present two DFR realizations with different nonlinear functions, achieving 10x power reduction and 5.3x throughput improvement while maintaining equal or better accuracy.

cs.AR↗

arXiv2023

A Foundation of $σ$-superadditive Measures -- a note on advancing Kalina measures -

The present paper attempts to modify the way of constructing a measure in the Alternative Set Theory setting originally devised by Martin Kalina. Introducing a system of cuts of rational numbers extended with some special ones, it is proved that the measure which is nondecreasing, nonnegative and "depending on the way of measurement" as same as Kalina's, but exhibits $σ$-superadditivity is obtained.

math.LO↗

arXiv2022

Unicellular LLT polynomials and twin of regular semisimple Hessenberg varieties

The solution of Shareshian-Wachs conjecture by Brosnan-Chow linked together the cohomology of regular semisimple Hessenberg varieties and graded chromatic symmetric functions on unit interval graphs. On the other hand, it is known that unicellular LLT polynomials have similar properties to graded chromatic symmetric functions. In this paper, we link together the unicellular LLT polynomials and twin of regular semisimple Hessenberg varieties introduced by Ayzenberg-Buchstaber. We prove their palindromicity from topological viewpoint. We also show that modules of a symmetric group generated by faces of a permutohedron are related to a shifted unicellular LLT polynomial and observe the $e$-positivity of shifted unicellular LLT polynomials, which is established by Alexandersson-Sulzgruber in general, for path graphs and complete graphs through the cohomology of the twins.

math.CO↗

arXiv2022

The second cohomology of regular semisimple Hessenberg varieties from GKM theory

We describe the second cohomology of a regular semisimple Hessenberg variety by generators and relations explicitly in terms of GKM theory. The cohomology of a regular semisimple Hessenberg variety becomes a module of a symmetric group $\mathfrak{S}_n$ by the dot action introduced by Tymoczko. As an application of our explicit description, we give a formula describing the isomorphism class of the second cohomology as an $\mathfrak{S}_n$-module. Our formula is not exactly the same as the known formula by Chow or Cho-Hong-Lee but they are equivalent. We also discuss its higher degree generalization.

math.AG↗

arXiv2022

Adaptive Outlier Detection for Power MOSFETs Based on Gaussian Process Regression

Outlier detection of semiconductor devices is important since manufacturing variation is inherently inevitable. In order to properly detect outliers, it is necessary to consider the discrepancy from underlying trend. Conventional methods are insufficient as they cannot track spatial changes of the trend}}. This study proposes an adaptive outlier detection using Gaussian process regression (GPR) with Student-t likelihood, which captures a gradual spatial change of characteristic variation. According to the credible interval of the GPR posterior distribution, the devices having excessively large deviations against the underlying trend are detected. The proposed methodology is validated by the experiments using a commercial SiC wafer and simulation.

eess.SY↗

arXiv2021

ENSEI: Efficient Secure Inference via Frequency-Domain Homomorphic Convolution for Privacy-Preserving Visual Recognition

In this work, we propose ENSEI, a secure inference (SI) framework based on the frequency-domain secure convolution (FDSC) protocol for the efficient execution of privacy-preserving visual recognition. Our observation is that, under the combination of homomorphic encryption and secret sharing, homomorphic convolution can be obliviously carried out in the frequency domain, significantly simplifying the related computations. We provide protocol designs and parameter derivations for number-theoretic transform (NTT) based FDSC. In the experiment, we thoroughly study the accuracy-efficiency trade-offs between time- and frequency-domain homomorphic convolution. With ENSEI, compared to the best known works, we achieve 5--11x online time reduction, up to 33x setup time reduction, and up to 10x reduction in the overall inference time. A further 33% of bandwidth reductions can be obtained on binary neural networks with only 1% of accuracy degradation on the CIFAR-10 dataset.

cs.CR↗

arXiv2021

Accelerating Parameter Extraction of Power MOSFET Models Using Automatic Differentiation

The extraction of the model parameters is as important as the development of compact model itself because simulation accuracy is fully determined by the accuracy of the parameters used. This study proposes an efficient model-parameter extraction method for compact models of power MOSFETs. The proposed method employs automatic differentiation (AD), which is extensively used for training artificial neural networks. In the proposed AD-based parameter extraction, gradient of all the model parameters is analytically calculated by forming a graph that facilitates the backward propagation of errors. Based on the calculated gradient, computationally intensive numerical differentiation is eliminated and the model parameters are efficiently optimized. Experiments are conducted to fit current and capacitance characteristics of commercially available silicon carbide MOSFET using power MOSFET models having 13 model parameters. Results demonstrated that the proposed method could successfully derive the model parameters 3.50x faster than a conventional numerical-differentiation method while achieving the equal accuracy.

eess.SY↗

arXiv2020

Virtual Secure Platform: A Five-Stage Pipeline Processor over TFHE

We present Virtual Secure Platform (VSP), the first comprehensive platform that implements a multi-opcode general-purpose sequential processor over Fully Homomorphic Encryption (FHE) for Secure Multi-Party Computation (SMPC). VSP protects both the data and functions on which the data are evaluated from the adversary in a secure computation offloading situation like cloud computing. We proposed a complete processor architecture with a five-stage pipeline, which improves the performance of the VSP by providing more parallelism in circuit evaluation. In addition, we also designed a custom Instruction Set Architecture (ISA) to reduce the gate count of our processor, along with an entire set of toolchains to ensure that arbitrary C programs can be compiled into our custom ISA. In order to speed up instruction evaluation over VSP, CMUX Memory based ROM and RAM constructions over FHE are also proposed. Our experiments show that both the pipelined architecture and the CMUX Memory technique are effective in improving the performance of the proposed processor. We provide an open-source implementation of VSP which achieves a per-instruction latency of less than 1 second. We demonstrate that compared to the best existing processor over FHE, our implementation runs nearly 1,600$\times$ faster.

cs.CR↗

arXiv2020

FedNNNN: Norm-Normalized Neural Network Aggregation for Fast and Accurate Federated Learning

Federated learning (FL) is a distributed learning protocol in which a server needs to aggregate a set of models learned some independent clients to proceed the learning process. At present, model averaging, known as FedAvg, is one of the most widely adapted aggregation techniques. However, it is known to yield the models with degraded prediction accuracy and slow convergence. In this work, we find out that averaging models from different clients significantly diminishes the norm of the update vectors, resulting in slow learning rate and low prediction accuracy. Therefore, we propose a new aggregation method called FedNNNN. Instead of simple model averaging, we adjust the norm of the update vector and introduce momentum control techniques to improve the aggregation effectiveness of FL. As a demonstration, we evaluate FedNNNN on multiple datasets and scenarios with different neural network models, and observe up to 5.4% accuracy improvement.

cs.LG↗

arXiv2020

Real Functions and its Differentiation in Alternative Set Theory

In the previous paper (Kiri Sakahara and Takashi Sato. Basic Topological Concepts and a Construction of Real Numbers in Alternative Set Theory. arXiv e-prints, arXiv:2005.04388, May 2020), the authors displayed basic topological concepts and a construction of a system of real number in alternative set theory (AST). The present paper is a continuation of that research providing additional treatments of real functions. The basic properties of differentiation in AST are preserved as in the conventional calculus.

math.LO↗

arXiv2020

BUNET: Blind Medical Image Segmentation Based on Secure UNET

The strict security requirements placed on medical records by various privacy regulations become major obstacles in the age of big data. To ensure efficient machine learning as a service schemes while protecting data confidentiality, in this work, we propose blind UNET (BUNET), a secure protocol that implements privacy-preserving medical image segmentation based on the UNET architecture. In BUNET, we efficiently utilize cryptographic primitives such as homomorphic encryption and garbled circuits (GC) to design a complete secure protocol for the UNET neural architecture. In addition, we perform extensive architectural search in reducing the computational bottleneck of GC-based secure activation protocols with high-dimensional input data. In the experiment, we thoroughly examine the parameter space of our protocol, and show that we can achieve up to 14x inference time reduction compared to the-state-of-the-art secure inference technique on a baseline architecture with negligible accuracy degradation.

cs.CV↗

arXiv2020

Current status of space gravitational wave antenna DECIGO and B-DECIGO

Deci-hertz Interferometer Gravitational Wave Observatory (DECIGO) is the future Japanese space mission with a frequency band of 0.1 Hz to 10 Hz. DECIGO aims at the detection of primordial gravitational waves, which could be produced during the inflationary period right after the birth of the universe. There are many other scientific objectives of DECIGO, including the direct measurement of the acceleration of the expansion of the universe, and reliable and accurate predictions of the timing and locations of neutron star/black hole binary coalescences. DECIGO consists of four clusters of observatories placed in the heliocentric orbit. Each cluster consists of three spacecraft, which form three Fabry-Perot Michelson interferometers with an arm length of 1,000 km. Three clusters of DECIGO will be placed far from each other, and the fourth cluster will be placed in the same position as one of the three clusters to obtain the correlation signals for the detection of the primordial gravitational waves. We plan to launch B-DECIGO, which is a scientific pathfinder of DECIGO, before DECIGO in the 2030s to demonstrate the technologies required for DECIGO, as well as to obtain fruitful scientific results to further expand the multi-messenger astronomy.

gr-qc↗

arXiv2020

Basic Topological Concepts and a Construction of Real Numbers in Alternative Set Theory

Alternative set theory (AST) may be suitable for the ones who try to capture objects or phenomenons with some kind of indefiniteness of a border. While AST provides various notions for advanced mathematical studies, correspondence of them to that of conventional ones are not fully developed. This paper presents basic topological concepts in AST, and shows their correspondence with those of conventional ones, and isomorphicity of a system of real numbers in AST to that of conventional ones.

math.LO↗

arXiv2020

NASS: Optimizing Secure Inference via Neural Architecture Search

Due to increasing privacy concerns, neural network (NN) based secure inference (SI) schemes that simultaneously hide the client inputs and server models attract major research interests. While existing works focused on developing secure protocols for NN-based SI, in this work, we take a different approach. We propose NASS, an integrated framework to search for tailored NN architectures designed specifically for SI. In particular, we propose to model cryptographic protocols as design elements with associated reward functions. The characterized models are then adopted in a joint optimization with predicted hyperparameters in identifying the best NN architectures that balance prediction accuracy and execution efficiency. In the experiment, it is demonstrated that we can achieve the best of both worlds by using NASS, where the prediction accuracy can be improved from 81.6% to 84.6%, while the inference runtime is reduced by 2x and communication bandwidth by 1.9x on the CIFAR-10 dataset.

cs.CR↗

arXiv2019

An Alternative Set Model of Cognitive Jump

When we enumerate numbers up to some specific value, or, even if we do not specify the number, we know at the same time that there are much greater numbers which should be reachable by the same enumeration, but indeed we also congnize them without practical enumeration. Namely, if we deem enumeration to be a way of reaching a number without any "jump", there is a "jump'' in our way of cognition of such greater numbers. In this article, making use of a set theoretical framework by Vopěnka (1979) (alternative set theory) which describes such structure, we attempt to shed light on an analogous sturucture in human and social phenomenon. As an example, we examine a problem of common knowledge in electronic mail game presented by Rubinstein (1989). We show an event comes to common knowledge by a "cognitive jump".

cs.GT↗

arXiv2019

GPU Based Parallel Ising Computing for Combinatorial Optimization Problems in VLSI Physical Design

In VLSI physical design, many algorithms require the solution of difficult combinatorial optimization problems such as max/min-cut, max-flow problems etc. Due to the vast number of elements typically found in this problem domain, these problems are computationally intractable leading to the use of approximate solutions. In this work, we explore the Ising spin glass model as a solution methodology for hard combinatorial optimization problems using the general purpose GPU (GPGPU). The Ising model is a mathematical model of ferromagnetism in statistical mechanics. Ising computing finds a minimum energy state for the Ising model which essentially corresponds to the expected optimal solution of the original problem. Many combinatorial optimization problems can be mapped into the Ising model. In our work, we focus on the max-cut problem as it is relevant to many VLSI physical design problems. Our method is inspired by the observation that Ising annealing process is very amenable to fine-grain massive parallel GPU computing. We will illustrate how the natural randomness of GPU thread scheduling can be exploited during the annealing process to create random update patterns and allow better GPU resource utilization. Furthermore, the proposed GPU-based Ising computing can handle any general Ising graph with arbitrary connections, which was shown to be difficult for existing FPGA and other hardware based implementation methods. Numerical results show that the proposed GPU Ising max-cut solver can deliver more than 2000X speedup over the CPU version of the algorithm on some large examples, which shows huge performance improvement for addressing many hard optimization algorithms for practical VLSI physical design.

physics.comp-ph↗

arXiv2016

Hessenberg varieties and hyperplane arrangements

Given a semisimple complex linear algebraic group $G$ and a lower ideal $I$ in positive roots of $G$, three objects arise: the ideal arrangement $\mathcal{A}_I$, the regular nilpotent Hessenberg variety $\mbox{Hess}(N,I)$, and the regular semisimple Hessenberg variety $\mbox{Hess}(S,I)$. We show that a certain graded ring derived from the logarithmic derivation module of $\mathcal{A}_I$ is isomorphic to $H^*(\mbox{Hess}(N,I))$ and $H^*(\mbox{Hess}(S,I))^W$, the invariants in $H^*(\mbox{Hess}(S,I))$ under an action of the Weyl group $W$ of $G$. This isomorphism is shown for general Lie type, and generalizes Borel's celebrated theorem showing that the coinvariant algebra of $W$ is isomorphic to the cohomology ring of the flag variety $G/B$. This surprising connection between Hessenberg varieties and hyperplane arrangements enables us to produce a number of interesting consequences. For instance, the surjectivity of the restriction map $H^*(G/B)\to H^*(\mbox{Hess}(N,I))$ announced by Dale Peterson and an affirmative answer to a conjecture of Sommers-Tymoczko are immediate consequences. We also give an explicit ring presentation of $H^*(\mbox{Hess}(N,I))$ in types $B$, $C$, and $G$. Such a presentation was already known in type $A$ or when $\mbox{Hess}(N,I)$ is the Peterson variety. Moreover, we find the volume polynomial of $\mbox{Hess}(N,I)$ and see that the hard Lefschetz property and the Hodge-Riemann relations hold for $\mbox{Hess}(N,I)$, despite the fact that it is a singular variety in general.

math.AG↗

arXiv2014

$p$-local stable splitting of quasitoric manifolds

We show a homotopy decomposition of $p$-localized suspension $ΣM_{(p)}$ of a quasitoric manifold $M$ by constructing power maps. As an application we investigate the $p$-localized suspension of the projection $π$ from the moment-angle complex onto $M$, from which we deduce its triviality for $p>\dim M/2$. We also discuss non-triviality of $π_{(p)}$ and $Σ^\inftyπ$.

math.AT↗