テクニカルレポート on 池田思朗

Faraday Tomography with Sparse Modeling

Thu, 01 Nov 2018 00:00:00 +0000

著者:

Kazunori Akiyama
Takuya Akahori
Yoshimitsu Miyashita
Shinsuke Ideguchi
Ryosuke Yamaguchi
Shiro Ikeda
Keitaro Takahashi

URL:

arXiv

Abstract:

Faraday tomography (or rotation measure synthesis) is a procedure to convert linear polarization spectra into the Faraday dispersion function, which provides us with unique information of magneto-ionic media along the line of sight. Mathematical formulation of Faraday tomography is similar to polarimetric imaging of radio interferometry, where many new methods have been actively developed and shown to outperform the standard CLEAN approaches. In this paper, we propose a sparse reconstruction technique to Faraday tomography. This technique is being developed for interferometric imaging and utilizes computationally less expensive convex regularization functions such as -norm and total variation (TV) or total squared variation (TSV). The proposed technique solves a convex optimization, and therefore its solution is determined uniquely regardless of the initial condition for given regularization parameters that can be optimized by data themselves. Using a physically-motivated model of turbulent galactic magnetized plasma, we demonstrate that the proposed technique outperforms RM-CLEAN and provides higher-fidelity reconstruction. The proposed technique would be a powerful tool in broadband polarimetry with the Square Kilometre Array (SKA) and its precursors.

Application of data science techniques to disentangle X-ray spectral variation of super-massive black holes

Wed, 01 Mar 2017 00:00:00 +0000

JAXA/ISAS, no. JAXA-RR-16-007

著者:

Sean Pike
Ken Ebisawa
Shiro Ikeda
Mikio Morii
Misaki Mizumoto
Eriko Kusunoki

URL:

DOI
arXiv

Abstract:

We apply three data science techniques, Nonnegative Matrix Factorization (NMF), Principal Component Analysis (PCA) and Independent Component Analysis (ICA), to simulated X-ray energy spectra of a particular class of super-massive black holes. Two competing physical models, one whose variable components are additive and the other whose variable components are multiplicative, are known to successfully describe X-ray spectral variation of these super-massive black holes, within accuracy of the contemporary observation. We hope to utilize these techniques to compare the viability of the models by probing the mathematical structure of the observed spectra, while comparing advantages and disadvantages of each technique. We find that PCA is best to determine the dimensionality of a dataset, while NMF is better suited for interpreting spectral components and comparing them in terms of the physical models in question. ICA is able to reconstruct the parameters responsible for spectral variation. In addition, we find that the results of these techniques are sufficiently different that applying them to observed data may be a useful test in comparing the accuracy of the two spectral models.

Supernova candidates discovered with Subaru/Hyper Suprime-Cam

Tue, 01 Dec 2015 00:00:00 +0000

The Astronomer’s Telegram, no. ATel#7927

著者:

Nozomu Tominaga
Tomoki Morokuma
Masaomi Tanaka
Naoki Yasuda
Hisanori Furusawa
Mikio Morii
Shiro Ikeda
Naonori Ueda
Junji Yamato
Katsuhiko Ishiguro
Yoshitaka Nakamura
Akihiro Yamanaka
Naoki Yoshida
Nao Suzuki
Ji-an Jiang
Takahiro Kato
Yuki Taniguchi
Takumi Shibata
Satoshi Miyazaki
Takashi J. Moriya
Junichi Noumaru
Kiaina Schubert
Tadafumi Takata

URL:

Link

Abstract:

We report the discovery of 10 supernova candidates from a transient survey with Subaru/Hyper Suprime-Cam (HSC). Our Subaru/HSC open-use observations were performed on 19 Aug 2015 UT, under poor weather condition with 1.1 – 1.5 arcsec seeing.

Capacity of a Single Neuron Channel

Wed, 01 Oct 2008 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 1076

著者:

Shiro Ikeda
Jonathan Manton

URL:

Link

Abstract:

The information transfer through a single neuron is a fundamental information processing in the brain and computing the information channel capacity is important to understand the information processing in the brain. The problem is difficult since the capacity depends on various issues, such as coding, characteristics of the communication channel and optimisation over input distributions. In this letter, two different models are considered. The temporal coding model of a neuron as a communication channel assumes the output is $\tau$ where $\tau$ is a gamma-distributed random variable corresponding to the inter-spike interval, that is, the time it takes for the neuron to fire once. The rate coding model is similar; the output is the actual rate of firing over a fixed period of time. Theoretical studies prove that the distribution of inputs, which achieves the channel capacity, is a discrete distribution with finite mass points for temporal and rate coding under a reasonable assumption. This allows us to compute numerically the capacity of a neuron. Numerical results are in a plausible range based on biological evidence to date.

Channel Estimation and Code Word Inference for Mobile Digital Satellite Broadcasting Reception

Wed, 01 Oct 2008 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 1077

著者:

Masatoshi Hamada
Shiro Ikeda

URL:

Link

Abstract:

This paper proposes a method of improving reception of digital satellite broadcasting in a moving vehicle. According to some studies, the antennas used for mobile reception will be smaller in the next generation and reception will be more difficult because of a fading multipath channel with delays in a low carrier-to-noise ratio. Commonly used approaches to reduce the inter symbol interference caused by a fading multipath channel with delays are pilot sequences and diversity reception. Digital satellite broadcasting, however, does not transmit pilot sequences for channel estimation and it is not possible to install multiple antennas in a vehicle. This paper does not propose any change to the broadcasting standards but discusses how to process currently available digital satellite signals to obtain better results. Our method does not rely on the pilot sequences or diversity reception, but consists of channel estimation and stochastic inference methods. For each task, two methods are proposed. The maximum likelihood estimation and higher order statistics matching methods are proposed for the estimation, and the marginal with the joint probability inference methods are proposed for the stochastic inference. The improvements were confirmed through experiments with numerical simulations and real data. The computational costs are also discussed for future implementation.

Motor planning as an optimization of command representation

Sat, 01 Sep 2007 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 1045

著者:

Shiro Ikeda
Yutaka Sakaguchi

URL:

Link

Abstract:

The fundamental problem of the motor neuroscience is to understand how humans make precise movements effortlessly. The problem seems difficult since there are infinite possible trajectories and the muscles are generally redundant. We discuss the problem from the viewpoint of motor command representation and show that a simple strategy can solve the problem with a two joints arm model. We also discuss the emergence of the muscle synergies, which may enable us to make natural motor behaviors with small degrees of freedom.

A bridge between boosting and a kernel machine

Fri, 01 Sep 2006 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 1006

著者:

Masanori Kawakita
Shiro Ikeda
Shinto Eguchi

URL:

Link

Abstract:

In this paper, boosting methods are studied from a viewpoint of kernel machines. This natural connection has already been revealed by defining a kernel function associated with the set of weak learners, which we call the WL kernel (Weak Learner kernel). We review this connection with respect to a kernel exponential family, and propose two important extensions of boosting methods for classification problems. First proposal is a new simple regularized boosting, which is confirmed to be valid through some experiments on real data. The other is a new simple kernel function from the investigation of the RKHS of decision stumps, which is one of the most widely-used weak learners. Several experiments confirm the efficiency and the validity of the proposed algorithm with the new kernel function.

Motor Planning and Sparse Motor Command Representation

Fri, 01 Sep 2006 00:00:00 +0000

Technical Report of University of Electro-Communications, no. UEC-IS-2006-1

著者:

Yutaka Sakaguchi
Shiro Ikeda

Abstract:

The present article proposes a novel computational approach to the motor planning. In the proposed approach, each motor command is represented as a linear combination of prefixed basis patterns, and the command for a given task is designed by minimizing a two-termed ``information representation criterion’’ which consists of a task optimization term and a parameter preference term. The result of a computer simulation with a single-joint reaching task confirmed that the proposed framework appropriately worked, together with showing that the resultant trajectory qualitatively replicated Fitts’ law.

Sparse representation and piece-wise linear kernel

Sat, 01 Oct 2005 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 962

著者:

Shiro Ikeda

URL:

Link

Abstract:

We propose a new type of kernel function, where feature space is explicitly given with a piece-wise linear mapping from the input space. This idea is inspired by sparse linear system analysis, where inputs are represented as a sparse linear combination of ``dictionary vectors.’’ This article gives the idea of such kernel function, and some preliminary experimental results.

Stochastic Reasoning, Free Energy and Information Geometry

Mon, 01 Sep 2003 00:00:00 +0000

The Institute of Statistical Mathematics, Research Memorandum, no. 890

著者:

Shiro Ikeda
Toshiyuki Tanaka
Shun-ichi Amari

URL:

Link

Abstract:

Belief propagation is a universal method of stochastic reasoning. It gives a good approximate solution, when it is applied to a stochastic model with loopy interactions. AI, statistical physical, and information geometrical methods have so far been used to analyze its performance separately. The present paper gives a unified framework to understand the relation underlying these concepts. In particular, the free energy and its relation to BP and CCCP is elucidated from the point of view of information geometry. We then propose a family of new algorithms. The stability of the algorithms is also analyzed, and methods of accelerating these algorithms are proposed.

Information Geometry of Turbo and Low-Density Parity-Check Codes

Wed, 01 Jan 2003 00:00:00 +0000

BSIS Technical Report, no. 03-2

著者:

Shiro Ikeda
Toshiyuki Tanaka
Shun-ichi Amari

Abstract:

Since the proposal of turbo codes in 1993, many studies have appeared on this simple and new type of codes which give a powerful and practical performance of error correction. Although experimental results strongly support the efficacy of turbo codes, further theoretical analysis is necessary, which is not straightforward. It is pointed out that the iterative decoding algorithm of turbo codes shares essentially similar ideas with low-density parity-check (LDPC) codes, with Pearl’s belief propagation algorithm applied to a cyclic belief diagram, and with the Bethe approximation in statistical physics. Therefore, the analysis of the turbo decoding algorithm will reveal the mystery of those similar iterative methods. In this paper, we recapture and extend the geometrical framework initiated by Richardson to the information geometrical framework of dual affine connections, focusing on both of the turbo and LDPC decoding algorithms. The framework helps our intuitive understanding of the algorithms and opens a new prospect of further analysis. We reveal some properties of these codes in the proposed framework, including the stability and error analysis. Based on the error analysis, we finally propose a correction term for improving the approximation.

Information Geometry and Mean Field Approximation: The $\alpha$-projection Approach

Mon, 01 May 2000 00:00:00 +0000

BSIS Technical Report, no. 00-6

著者:

Shun-ichi Amari
Shiro Ikeda
Hidetoshi Shimokawa

Abstract:

Information geometry is applied to mean field approximation for elucidating its properties in the spin glass model or the Boltzmann machine. The $\alpha$-divergence is used for approximation, where $\alpha$-geodesic projection plays an important role. The naive mean field approximation and TAP approximation are studied from the point of view of information geometry, which treats the intrinsic geometric structures of a family of probability distributions. The bifurcation of the $\alpha$-projection is studied, at which the uniqueness of the $\alpha$-approximation is broken.

ICA for noisy neurobiological data

Sat, 01 Apr 2000 00:00:00 +0000

BSIS Technical Report, no. 00-2

著者:

Shiro Ikeda
Keisuke Toyama

Abstract:

ICA (independent component analysis) is a new, simple and powerful idea for analyzing multi-variant data. One of the successful applications is neurobiological data analysis such as EEG (electroencephalography), MRI (magnetic resonance imaging), and MEG (magnetoencephalography). But there remain a lot of problems. In most cases, neurobiological data contain a lot of sensory noise, and the number of independent components is unknown. In this article, we discuss an approach to separate noise-contaminated data without knowing the number of independent components. A well-known two stage approach to ICA is to pre-process the data by PCA (principal component analysis), and then the necessary rotation matrix is estimated. Since PCA does not work well for noisy data, we implement a factor analysis model for pre-processing. In the new pre-processing, the number of the sources and the amount of the sensory noise are estimated. After the pre-processing, the rotation matrix is estimated using an ICA method. Through the experiments with MEG data, we show this approach is effective.

An Approach to Blind Source Separation Based on Temporal Structure of Speech Signals

Wed, 01 Apr 1998 00:00:00 +0000

BSIS Technical Report, no. 98-2

著者:

Noboru Murata
Shiro Ikeda
Andreas Ziehe

キーワード:

sound separation
ICA
signal processing

Abstract:

In this paper we introduce a new technique for blind source separation of speech signals. We focus on the temporal structure of the signals in contrast to most other major approaches to this problem. The idea is to apply the decorrelation method proposed by Molgedey and Schuster in the time-frequency domain. We show some results of experiments with both artificially controlled data and speech data recorded in the real environment.

テクニカルレポート on 池田 思朗

Faraday Tomography with Sparse Modeling

著者:

URL:

Abstract:

Application of data science techniques to disentangle X-ray spectral variation of super-massive black holes

著者:

URL:

Abstract:

Supernova candidates discovered with Subaru/Hyper Suprime-Cam

著者:

URL:

Abstract:

Capacity of a Single Neuron Channel

著者:

URL:

Abstract:

Channel Estimation and Code Word Inference for Mobile Digital Satellite Broadcasting Reception

著者:

URL:

Abstract:

Motor planning as an optimization of command representation

著者:

URL:

Abstract:

A bridge between boosting and a kernel machine

著者:

URL:

Abstract:

Motor Planning and Sparse Motor Command Representation

著者:

Abstract:

Sparse representation and piece-wise linear kernel

著者:

URL:

Abstract:

Stochastic Reasoning, Free Energy and Information Geometry

著者:

URL:

Abstract:

Information Geometry of Turbo and Low-Density Parity-Check Codes

著者:

Abstract:

Information Geometry and Mean Field Approximation: The $\alpha$-projection Approach

著者:

Abstract:

ICA for noisy neurobiological data

著者:

Abstract:

An Approach to Blind Source Separation Based on Temporal Structure of Speech Signals

著者:

キーワード:

Abstract:

テクニカルレポート on 池田思朗