This paper presents a novel Wiener-type recurrent neural network with the observer/Kalman filter identification (OKID) algorithm for unknown dynamic nonlinear system identification. The proposed Wiener-type recurrent network resembles the conventional Wiener model that consists of a dynamic linear subsystem cascaded with a static nonlinear subsystem. The novelties of our approach include: (1) t...

The diagnosing of disease using speckled ultrasound image is a very difficult task for doctors. The speckle noise not only hinders the visual information but also affect the segmentation process used in ultrasound image processing. Thus speckle noise suppression is necessary pre-processing task in order to maintain the diagnostic potential of ultrasound imaging. This paper proposes an efficient...

The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) ...

Originaly, a filter is a physical device for removing unwanted components of mixtures (gas, liquid, solid). In the area of telecommunications, signals are mixtures of different frequencies, and the term of filter is used to describe the attenuation of the unwanted frequencies. Since 1940, the concept of a filter was extended to the separation of signals from noise. With Kalman filter, the meani...

The Wiener filter solves the Wiener-Hopf equation and may be approximated by the Multi-Stage Nested Wiener Filter (MSNWF) which lies in the Krylov subspace of the covariance matrix of the observation and the crosscorrelation vector between the observation and the desired signal. Moreover, since the covariance matrix is Hermitian, the Lanczos algorithm can be used to compute the Krylov subspace ...

The observables of the GOCE three-axis gradiometer are taken in time, along the orbit and, in the timewise proposal, filtered to stay in the measurement bandwidth, i.e. in the frequency interval between 0.005 and 0.1 Hz. As a consequence, the resulting “observable” is a convolution of the original data stream with a time-wise weighting kernel. In other words, we cannot assume that the observati...

The multistage Wiener filter (MWF) outperforms the full rank Wiener filter in low sample support environments. However, the MWF adaptive process should be stopped at an optimum stage to get the best performance. There are two methods to stop the MWF adaptive process. One method is to calculate until the final full-stage, and the second method is to terminate at r-stage less than full-stage. The...