We present an order-recursive lattice algorithm for H^∞ adaptive filtering. The standard H^∞ filter algorithm is expressed in an alternative formula that is closed in a relaxed sense. On the basis of this formula, the algorithm is reformulated as an indefinite least-squares problem that can be solved by a two-dimensional block recursive least-squares (RLS) algorithm. Then an H^∞ lattice filter is derived by transforming the block RLS algorithm into its multivariable least-squares lattice form. The resulting computational complexity is still proportional to the filter order, although 2 x 2 matrix computations are involved. The equivalence between the standard H^∞ filter and its lattice version is verified by numerical experiments.

抄録全体を表示

In this paper, we propose computational algorithms for a pseudo-formal linearization method for nonlinear dynamic systems and a nonlinear observer for nonlinear scalar measurement systems using Chebyshev interpolation. This pseudo-formal linearization method transforms a nonlinear autonomous dynamic system into an augmented linear one with respect to a linearization function that consists of polynomials of state variables. When linearizing, Chebyshev interpolation is exploited so that a computational algorithm is implemented. As an application of this method, a computational algorithm for a nonlinear observer is proposed. Numerical experiments indicate that the performance of the proposed method is superior to that of the previous method.

抄録全体を表示

Detection of disaster-stricken areas using synthetic aperture radar (SAR) images is important in countries and regions with heavy rain and earthquakes. Although it is important to immediately find disaster-stricken areas when a disaster occurs, it takes time to read SAR images and also needs experience and expertise. Therefore, machine learning, especially deep learning, is expected to be applied to the classification of disaster-stricken areas. Classification using deep learning is often executed on patch images of local areas. However, patch-based classification would miss information on the surrounding areas such as topographic features. In this study, a convolutional neural network (CNN) is applied to the classification of SAR images using the following techniques. When making the images input to a CNN, two multichannel image generation methods, i.e, a zero-padding method and map-concatenation method, are used, where the target areas to be classified and their surrounding areas are combined to form multichannel images. In the experiments, the zero-padding method and map-concatenation method are evaluated by the classification performance of SAR images that cover the northern Kyushu area in Japan, where large-scale landslides due to heavy rain occurred in 2017. Through the experiments, we clarify the appropriate CNN structures with multichannel information for landslide classification.

抄録全体を表示

The basic particles of quanta are quarks and leptons. Quarks are combined with each other and form neutrons and protons. Neutrons and protons are combined with each other and form quanta called atomic nuclei, which are considered to vibrate. This vibration is considered to be expressed by a wave equation. In this session, we show that the one-dimensional crystal described in Session 5 satisfies a quantized wave equation. Moreover, the one-dimensional crystal is equivalent to an LC ladder circuit, which consists of a coil and a capacitor, as also described in Session 5. Therefore, the applicability of circuit theory to atomic nuclei is shown by assuming that atomic nuclei satisfy the quantized wave equation as LC ladder circuits.

抄録全体を表示