• The KIPS Transactions:PartB
• /
• v.8B no.6
• /
• pp.699-704
• /
• 2001

ABSTRACT In this paper, a new hybrid modular recurrent TDNN (time-delay neural network)-HMM (hidden Markov model) architecture for speech recognition has been studied. In TDNN, the recognition rate could be increased if the signal window is extended. To obtain this effect in the neural network, a high-level memory generated through a feedback within the first hidden layer of the neural network unit has been used. To increase the ability to deal with the temporal structure of phonemic features, the input layer of the network has been divided into multiple states in time sequence and has feature detector for each states. To expand the network from small recognition task to the full speech recognition system, modular construction method has been also used. Furthermore, the neural network and HMM are integrated by feeding output vectors from the neural network to HMM, and a new parameter smoothing method which can be applied to this hybrid system has been suggested.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.2
• /
• pp.78.1-78.1
• /
• 2019

There are a huge number of faint objects that have not been observed due to the lack of large and deep surveys. In this study, we demonstrate that a deep learning approach can produce a better quality deep image from a single pass imaging so that could be an alternative of conventional image stacking technique or the expensive large and deep surveys. Using data from the Sloan Digital Sky Survey (SDSS) stripe 82 which provide repeatedly scanned imaging data, a training data set is constructed: g-, r-, and i-band images of single pass data as an input and r-band co-added image as a target. Out of 151 SDSS fields that have been repeatedly scanned 34 times, 120 fields were used for training and 31 fields for validation. The size of a frame selected for the training is 1k by 1k pixel scale. To avoid possible problems caused by the small number of training sets, frames are randomly selected within that field each iteration of training. Every 5000 iterations of training, the performance were evaluated with RMSE, peak signal-to-noise ratio which is given on logarithmic scale, structural symmetry index (SSIM) and difference in SSIM. We continued the training until a GAN model with the best performance is found. We apply the best GAN-model to NGC0941 located in SDSS stripe 82. By comparing the radial surface brightness and photometry error of images, we found the possibility that this technique could generate a deep image with statistics close to the stacked image from a single-pass image.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.19 no.10
• /
• pp.2028-2042
• /
• 1994

Properly defining signal and noise in a self-organizing system like ART(Adaptive Resonance Theory) neural network model raises a number of subtle issues. Pattern context must enter the definition so that input features, treated as irrelevant noise when they are embedded in a given input pattern, may be treated as informative signals when they are embedded in a different input pattern. The ATR automatically self-scales their computational units to embody context and learning dependent definitions of a signal and noise and there is no problem in categorizing input pattern that have features similar in nature. However, when we have imput patterns that have features that are different in size and nature, the use of only one vigilance parameter is not enough to differentiate a signal from noise for a good categorization. For example, if the value fo vigilance parameter is large, then noise may be processed as an informative signal and unnecessary categories are generated: and if the value of vigilance parameter is small, an informative signal may be ignored and treated as noise. Hence it is no easy to achieve a good pattern categorization. To overcome such problems, a Coupled-ART neural network capable of modularized categorization of patterns is proposed. The Coupled-ART has two layer of tightly coupled modules. the upper and the lower. The lower layer processes the global features of a pattern and the structural features, separately in parallel. The upper layer combines the categorized outputs from the lower layer and categorizes the combined output, Hence, due to the modularized categorization of patterns, the Coupled-ART classifies patterns more efficiently than the ART1 model.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.2
• /
• pp.307-314
• /
• 2011

A new predistortion technique for dynamic spectrum allocation systems such as cognitive radio (CR) is proposed. The system model considered in this paper occupies a small band at a time, but the center frequency can be changed in the wide range of frequency. In this scenario. the front-end filter may not eliminate the harmonics of the power amplifier (PA) output. The proposed PD reduces the spectral regrowth of the fundamental signal at the carrier frequency (${\omega}_0$) and removes the harmonics ($2{\omega}_0$, $3{\omega}_0$, ...) at the same time. The proposed PD structure is composed of multiple predistorters (PDs) centered at integer multiples of ${\omega}_0$. The PD at ${\omega}_0$ is for removing spectral regrowth of the fundamental signal, and the others are for harmonic reduction. In the proposed PD structure, parameters of PDs are found jointly. Simulation results show that the spectral regrowth can be reduced by 20dB, and the 2nd and 3rd harmonics can be reduced down to -70dB from the power of the fundamental signal.

• Journal of Sensor Science and Technology
• /
• v.4 no.4
• /
• pp.29-40
• /
• 1995

In this paper, the noise equivalent cicuits that is necessary to the formulation of D.C. and noise characteristics, residual component and input capacitance so as to analyze on the noise factors of the SIT is proposed. The simplest noise equivalent circuit is the model representing the mechanism of the SIT and the measured values in this model were found as small as the values of the shot-noise. In the source resistance inserted equivalent circuit is conformed that the shot-noise will be reduced by the negative-feedback effect of the source resistance. In oder to analyze the correct noise reduction factor, I proposed the equivalent circuit which the formulas of the source and drain resistance was induced. In the experiment which affirm the equivalent circuits, the influence of the signal source resistance and output load resistance on the residual component is small and the residual component can be expressed by the equivalent input noise resistance. Moreover, the input capacitance is 13.6 pF when the load resistance is $0{\Omega}$ and the capacitance which does not concern with the SIT operation directly, that is, gate wire etc, is 10pF or so.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_1
• /
• pp.793-806
• /
• 2020

In this study,for YSU (Yonsei University), MYJ(Mellor-Yamada-Janjic), ACM2 (Asymmetric Convective Model), and BouLac (Bougeault-Lacarrere) PBL schemes, numerical experiments were performed for the case period (June 26-30, 2014). The PBLH calculated by using the backscatter signal produced by the mobile vehicle-mounted lidar system (LIVE) and the PBLH calculated by the prediction of each PBL schemes of WRF were compared and analyzed. In general, the experiments using the non-local schemes showed a higher correlation than the local schemes for lidar observation. The standard deviation of the PBLH difference for daylight hours was small in the order of YSU (≈0.39 km), BouLac (≈0.45 km), ACM2 (≈0.47 km), MYJ (≈0.53 km) PBL schemes. In the RMSE comparison for the case period, the YSU PBL scheme was found to have the highest precision. The meteorological lider mounted on the vehicle is expected to provide guidance for the analysis of the planetary boundary layer in a numerical model under various weather conditions.

• The Bulletin of The Korean Astronomical Society
• /
• v.41 no.2
• /
• pp.64.3-65
• /
• 2016

The NISS (Near-infrared Imaging Spectrometer for Star formation history) is the near-infrared instrument optimized to the Next Generation of small satellite series (NEXTSat). The capability of both imaging and low spectral resolution spectroscopy in the near-infrared range is a unique function of the NISS. The major scientific mission is to study the cosmic star formation history in local and distant universe. For those purposes, the main observational targets are nearby galaxies, galaxy clusters, star-forming regions and low background regions. The off-axis optical design is optimized to have a wide field of view ($2deg.{\times}2deg.$) as well as the wide wavelength range from 0.95 to $3.8{\mu}m$. Two linear variable filters are used to realize the imaging spectroscopy with the spectral resolution of ~20. The mechanical structure is considered to endure the launching condition as well as the space environment. The compact dewar is confirmed to operate the infrared detector as well as filters at 80K stage. The electronics is tested to obtain and process the signal from infrared sensor and to communicate with the satellite. After the test and calibration of the engineering qualification model (EQM), the flight model of the NSS is assembled and integrated into the satellite. To verify operations of the satellite in space, the space environment tests such as the vibration, shock and thermal-vacuum test were performed. Here, we report the test results of the flight model of the NISS.

• Journal of Power Electronics
• /
• v.11 no.2
• /
• pp.194-201
• /
• 2011

The air flow supplied to a fuel cell system is one of the most significant factors in determining fuel efficiency. The conventional method of controlling the air flow is to fix the oxygen supply at an estimated constant rate for optimal efficiency. However, the actual optimal point can deviated from the pre-set value due to temperature, load conditions and so on. In this paper, the maximum efficiency point tracking (MEPT) algorithm is proposed for finding the optimal air supply rate in real time to maximize the net-power generation of fuel cell systems. The fixed step MEPT algorithm has slow dynamics, thus it affects the overall efficiency. As a result, the variable step MEPT algorithm is proposed to compensate for this problem instead of a fixed one. The complete small signal model of a PEM Fuel cell system is developed to perform a stability analysis and to present a design guideline. For a design example, a 1kW PEM fuel cell system with a DSP 56F807 (Motorola Inc) was built and tested using the proposed MEPT algorithm. This control algorithm is very effective for a soft current change load like a grid connected system or a hybrid electric vehicle system with a secondary energy source.

• Journal of Electrical Engineering and Technology
• /
• v.10 no.5
• /
• pp.2155-2161
• /
• 2015

This study suggests an optical method to measure cardinal of gait (step width, step length, and stride length) with position sensitive detector (PSD). The effect of reflector’s shape (flat and cylinder) on the PSD output voltage was examined for the application of the suggested system to real situations with a curved shape reflector (e.g. shoes). Various mathematical models were evaluated to find the optimal equation for the distance measurement. Considering the effect of shape on detected signal, the inverse polynomial model was developed. The suggested method is simple to operate, low in cost, small in size, and can evaluate gait parameters in real time. This method is expected to be useful in the field of rehabilitation and sport science

• Journal of Power Electronics
• /
• v.15 no.6
• /
• pp.1609-1618
• /
• 2015

DC microgrids are considered as prospective systems because of their easy connection of distributed energy resources (DERs) and electric vehicles (EVs), reduction of conversion loss between dc output sources and loads, lack of reactive power issues, etc. These features make them very suitable for future industrial and commercial buildings' power systems. In addition, the bipolar-type dc system structure is more popular, because it provides two voltage levels for different power converters and loads. To keep voltage balanced in such a dc system, a bidirectional dual buck-boost voltage balancer with direct coupling is introduced based on P-cell and N-cell concepts. This results in greatly enhanced system reliability thanks to no shoot-through problems and lower switching losses with the help of power MOSFETs. In order to increase system efficiency and reliability, a novel burst-mode control strategy is proposed for the dual buck-boost voltage balancer. The basic operating principle, the current relations, and a small-signal model of the voltage balancer are analyzed under the burst-mode control scheme in detail. Finally, simulation experiments are performed and a laboratory unit with a 5kW unbalanced ability is constructed to verify the viability of the bidirectional dual buck-boost voltage balancer under the proposed burst-mode control scheme in low-voltage bipolar-type dc microgrids.