| 1. |
- Navakauskas, Dalius, et al.
(författare)
-
Multirate Implementation Scheme for Restoration of Voiced Speech Signals
- 2003
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- Recently iterative procedure for the restoration of speech signals when prosodic elements: stress and accent, of comparatively long duration are missing was developed. Alternatively, it could be cast in a signal generation framework. Basing on that view the paper presents the efficient implementation scheme for the restoration of voiced speech signals. It enjoys parallel order of multirate processing utilizing interpolation and decimation filters parameterized by specific to problem coefficients.Presented simulation results confirm feasibility of developed implementation.
|
|
| 2. |
- Navakauskas, Dalius
(författare)
-
Speeding up the Training of Lattice-Ladder Multilayer Perceptrons
- 2002
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- A lattice-ladder multilayer perceptron (LLMLP) is an appealing structure for advanced signal processing in a sense that it is nonlinear, possesses infinite impulse response and stability monitoring of it during training is simple. However, even moderate implementation of LLMLP training hinders the fact that a lot of storage and computation power must be allocated. In this paper we deal with the problem of computational efficiency of LLMLP training algorithms that are based on computation of gradients, e.g., backpropagation, conjugate-gradient or Levenberg-Marquardt.The paper aims to explore most computationally demanding calculations---computation of gradients for lattice (rotation)parameters. Here we find and propose to use for training of several LLMLP architectures a simplest in terms of storage and number of delay elements computation of exact gradients, assuming that the coefficients of the lattice-ladder filter are held stationary.
|
|
| 3. |
- Navakauskas, Dalius
(författare)
-
Training Algorithm for Extra Reduced Size Lattice-Ladder Multilayer Perceptrons
- 2003
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- A quick gradient training algorithm for a specific neural network structure called an extra reduced size lattice-ladder multilayer perceptron is introduced. Presented derivation of the algorithm utilizes recently found by author simplest way of exact computation of gradients for rotation parameters of lattice-ladder filter. Developed neural network training algorithm is optimal in terms of minimal number of constants, multiplication and addition operations, while the regularity of the structure is also preserved.
|
|
| 4. |
- Pupeikis, Rimantas, et al.
(författare)
-
Identification of Wiener Systems with Hard and Discontinuous Nonlinearities
- 2003
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- The problem of identification of Wiener systems with special types of hard and discontinuous nonlinearities in the presence of process and measurements noises in observations to be processed has been considered. It is shown, that the original problem could be reduced to the problem of determination of the subsystem from the auxiliary network of subsystems, equivalent to the true linear system (linear part of the Wiener system). A technique based on the ordinary least squares, to be used in a case of missing data, and on the expectation maximization algorithm is proposed here. The results of numerical simulation of the discrete-time Wiener systems with various hard and discontinuous nonlinearities by computer are given.
|
|
| 5. |
- Pupeikis, Rimantas, et al.
(författare)
-
Recursive Parameter Estimation Using Closed-Loop Observations
- 2003
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- The aim of the given paper is development of a joint input-output approach in the case of an additive correlated noise acting on the output of the open-loop system. Here the ordinary prediction error method is applied to solve the closed-loop identification problem by processing observations. In the case of the known regulator, the two-stage method, which belongs to the ordinary joint input-output approach, reduces to the one-stage method. In such a case, the open-loop system could be easily determined after some extended rational transfer function has been identified. In the case of the unknown regulator, the estimate of the extended transfer function is used to generate an auxiliary input. The form of an additive noise filter, that ensures the minimal value of the mean square criterion, is determined. The results of numerical simulation and identification of the open-loop system by computer, using the two-stage method and closed-loop observations are given.
|
|
| 6. |
- Usinskas, Andrius, et al.
(författare)
-
Segmentation of Ischemic Stroke in CT Images of Human Brain
- 2003
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- The paper deals with the problem of segmentation of ischemic stroke in the human brain computer tomography images. Carried out thorough analysis of stroke regions in images shows up several useful features that can be used in early but rough image pre-processing stages. Based on these features a procedure for segmentation of ischemic stroke regions from 2D~images into 3D~shape is developed, presented and proved by experimentation.It can be used for evaluation of stroke volume and support the decision making about patient disability.
|
|
