• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.6
• /
• pp.922-931
• /
• 1995

In this paper, we propose a modified error function to improve the EBP(Error Back-Propagation) algorithm of Multi-Layer Perceptrons. Using the modified error function, the output node of MLP generates a strong error signal in the case that the output node is far from the desired value, and generates a weak error signal in the opposite case. This accelerates the learning speed of EBP algorothm in the initial stage and prevents overspecialization for training patterns in the final stage. The effectiveness of our modification is verified through the simulation of handwritten digit recognition.

• Proceedings of the IEEK Conference
• /
• 2002.07b
• /
• pp.739-742
• /
• 2002

This article proposes a Four-Quadrant Analog Multiplier (4-QAM) applying switched-capacitor and pulse-width amplitude modulation (PWAM) principles. The features of the presented circuit are that it can function as analog multiplier with a wide dynamic range of input signal and no disturbing from deviation of carrier frequency of PWM signal. In addition, the circuit detail is simpler than that of the previously proposed circuits. It is then easy and applicable for employing it into Integrated Circuit (IC) realization to especially operate in low-frequency and low-power applications. The experimental results granted are in correspondence to the theoretical analysis.

• Proceedings of the IEEK Conference
• /
• 1999.11a
• /
• pp.43-46
• /
• 1999

Blind equalizers recover the transmitted data using signal's statistical characteristics only. Because of its computational simplicity and fast convergence rate, CMA is widely used in practice. Blind equalizers, however, converge much slowly than conventional equalizers which use the training signals. In order to improve the convergence rate, many modified blind equalization algorithms have been proposed. Among those, Normalized CMA (NCMA) was applied to increase the convergence rate by using the large step size. Unfortunately it can only be applied for the constant modulus signal constellation scheme. this paper, we propose the Selective NCMA (SNCMA) that improve the convergence rate of blind equalization algorithms by using NCMA for non-constant modulus signalling method such as QAM constellation. We achieved fast start-up convergence rate and reduced steady-state residual error.

• Proceedings of the IEEK Conference
• /
• 2003.07a
• /
• pp.35-38
• /
• 2003

We propose a switched beamforming to improve the reception performance of VSB system in severe Rayleigh fading channel. The VSB system has only about 3% of known training signal for the receiver in a data field and the reception performance of VSB receiver is degraded significantly when there are near-0 dB ghosts in receiving signal. The switched beamforming forms several beams in different directions and selects only one beam among them. For the selection of a beam with best channel condition for VSB equalizer, we extracted the channel profiles in time domain for all the beams by correlating the PN511 sequence in VSB field sync and selected optimal beam by comparing the channel profiles. The simulation results show that the proposed method improves the reception performance of VSB system in severe Rayleigh channel.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.27 no.11
• /
• pp.1864-1872
• /
• 2003

Hidden Markov Model(HMM) has been widely used in speech recognition, however, its use in machine condition monitoring has been very limited despite its good potential. In this paper, HMM is used to recognize rotor fault pattern. First, we set up rotor kit under unbalance and oil whirl conditions. Time signals of two failure conditions were sampled and translated to auto power spectrums. Using filter bank, feature vectors were calculated from these auto power spectrums. Next, continuous HMM and discrete HMM were trained with scaled forward/backward variables and diagonal covariance matrix. Finally, each HMM was applied to all sampled data to prove fault recognition ability. It was found that HMM has good recognition ability despite of small number of training data set in rotor fault pattern recognition.

• Journal of the Institute of Convergence Signal Processing
• /
• v.15 no.2
• /
• pp.37-41
• /
• 2014

The vector Taylor series (VTS) based method usually employs clean speech Hidden Markov Models (HMMs) when compensating speech feature vectors or adapting the parameters of trained HMMs. It is well-known that noisy speech HMMs trained by the Multi-condition TRaining (MTR) and the Multi-Model-based Speech Recognition framework (MMSR) method perform better than the clean speech HMM in noisy speech recognition. In this paper, we propose a method to use the noise-adapted HMMs in the VTS-based speech feature compensation method. We derived a novel mathematical relation between the train and the test noisy speech feature vector in the log-spectrum domain and the VTS is used to estimate the statistics of the test noisy speech. An iterative EM algorithm is used to estimate train noisy speech from the test noisy speech along with noise parameters. The proposed method was applied to the noise-adapted HMMs trained by the MTR and MMSR and could reduce the relative word error rate significantly in the noisy speech recognition experiments on the Aurora 2 database.

• Proceedings of the KOSOMBE Conference
• /
• v.1998 no.11
• /
• pp.207-208
• /
• 1998

Classification of extracellularly recorded action potential into each unit is an important procedure for further analysis of spike trains as point process. We utilize feedforward neural network structures, multilayer perceptron and radial basis function network to implement spike classifier. For the efficient training of classifiers, nonlinear energy operator that can trace the instantaneous frequency as well as the amplitude of the input signal is used. Trained classifiers shows successful operation, up to 90% correct classification was possible under 1.2 of signal-to-noise ratio.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2001.06a
• /
• pp.221-224
• /
• 2001

In this paper we propose a shift-invariant uHMT estimation for wavelet-based image denoising. The proposed estimation have just nine meta-parameter (independent of the size of the image and the number of wavelet scales) and requires no kinds of training. Also it solve visual artifacts resulted in the lack of shift-invariance in the DWT. The experimental results show that the proposed estimation is more effective than the other wavelet-based denoising by 0.5-ldB (PSNR) and allows an Ο(nlog n) in terms of performance speed.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.10A
• /
• pp.1639-1644
• /
• 2001

Two decision feedback equalizer structures employing recurrent neural network (RNN) used for non-linear channels with severe intersymbol interference (ISI) and non-linear distortion are proposed in this paper, which skillfully put the traditional decision feedback structure for linear channels equalization into RNN, replace decision feedback signal with training signal in the learning process and adaptively adjust the learning step. Simulative results of the first type of two new equalizer structures have shown that it has better equalization performances than traditional recurrent neural network equalizer (RNNE) under the same condition.

• ETRI Journal
• /
• v.30 no.4
• /
• pp.546-554
• /
• 2008

Phase-locked loops (PLLs) are among the most important mixed-signal building blocks of modern communication and control circuits, where they are used for frequency and phase synchronization, modulation, and demodulation as well as frequency synthesis. The growing popularity of PLLs has increased the need to test these devices during prototyping and production. The problem of distinguishing and classifying the responses of analog integrated circuits containing catastrophic faults has aroused recent interest. This is because most analog and mixed signal circuits are tested by their functionality, which is both time consuming and expensive. The problem is made more difficult when parametric variations are taken into account. Hence, statistical methods and techniques can be employed to automate fault classification. As a possible solution, we use the back propagation neural network (BPNN) to classify the faults in the designed charge-pump PLL. In order to classify the faults, the BPNN was trained with various training algorithms and their performance for the test structure was analyzed. The proposed method of fault classification gave fault coverage of 99.58%.