• Journal of KIISE:Software and Applications
• /
• v.26 no.12
• /
• pp.1383-1392
• /
• 1999

이미 학습된 다층퍼셉트론 신경망을 디지털 VLSI 기술을 사용하여 하드웨어로 구현할 경우 신경망의 가중치 및 뉴런 출력들을 양자화해야 하는 문제가 발생한다. 이러한 신경망 변수들의 양자화는 결과적으로 주어진 입력에 대한 신경망의 최종 출력에서의 왜곡을 초래한다. 본 논문에서는 먼저 이러한 양자화로 인한 신경망 출력에서의 왜곡을 통계적으로 분석하였다. 분석 결과에 의하면 입력패턴 각 성분의 제곱들의 합과 가중치의 크기들이 양자화 영향에 주로 기여하는 것으로 나타났다. 이러한 분석 결과를 이용하여 양자화를 위한 정밀도가 주어졌을 때, 양자화 영향이 최소화된 다층퍼셉트론 신경망을 설계하는 방법을 제시하였다. 그리고 제안된 방법에 의해 얻은 신경망과 오류역전파 학습방법에 의하여 얻은 신경망의 성능을 비교함으로써 제안된 방법의 효율성을 입증하였다. 실험결과는 낮은 양자화 정밀도에서도 제안된 방법이 더 좋은 성능을 보였다.Abstract When we implement a multilayer perceptron with the digital VLSI technology, we generally have to quantize the weights and the neuron outputs. These quantizations eventually cause distortion in the output of the network for a given input. In this paper first we made a statistical analysis about the effect caused by the quantization on the output of the network. The analysis revealed that the sum of the squared input components and the sizes of the weights are the major factors which contribute to the quantization effect. We present a design method for an MLP which minimizes the quantization effect when the precision of the quantization is given. In order to show the effectiveness of the proposed method, we developed a network by our method and compared it with the one developed by the regular backpropagation. We could confirm that the network developed by our method performs better even with a low precision of the quantization.

• Journal of Internet Computing and Services
• /
• v.9 no.5
• /
• pp.47-58
• /
• 2008

Charts can propagate information more effectively by visualizing various types of information. Due to this reason, many web developers often use charts when they display information on the web. However, using a chart requires a special dedicated software to install and it is difficult to share the finished charts since they are usually in raster format. The raster format images are fixed in size, therefore, experience image distortion problem when their sizes are changed. In this paper, we propose a web-based chart generating system using CML(Chart Markup Language) to solve the current compatibility and sharing problems. The proposed chart generating system first analyzes XML, Text, or HTML documents to extract chart information in the documents, and then it converts collected chart information to CML. The converted CML documents are displayed on the web browser using a vector method. The vector method has an advantage of protecting its images from distortion even when image sizes are scaled. The CML, proposed in this paper, is a chart markup language based on XML for modeling charts Information. If CML is used for presenting a chart in a web, it makes easier to share and to convert the chart information.

• The Journal of the Acoustical Society of Korea
• /
• v.8 no.5
• /
• pp.33-50
• /
• 1989

In this paper, musical tone, especially instrument tones are synthesized using nonlinear distortion technique. Nonlinear distortion is very simple but versatile method when you synthesize musical instrument tones. It basically consists of one sine oscillator and amplifier which makes distortion to Input wave. Output wave has many harmonics that can be controlled by varying shaping function, which is the transfer function of nonlinear amplifier. Shaping function Is obtained from the analyzed harmonic amplitude data. Given harmonics amplitudes, Chebyshev polynomial is used to produce the shaping function that exactly makes the given harmonics at steady state. We contructed non -real time nonlinear distortion synthesizer program running at IBM-PC. To quantify the satis faction of synthesised tones, listening test is carried out, and the result is presented.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.40 no.11
• /
• pp.37-44
• /
• 2003

A distortion suppression technology for employing multiple inputs in 3n＋1 type current mode Max circuit is proposed using the adjustment of transconductance. If the number of input blocks of the current mode Max circuit increases, the high frequency distortion in the output signal grows. In this paper, it has been disclosed that the distortion in the multiple input Max circuit is proportional to such accumulated parasitic capacitance, to the derivative of the output signal and also to tile inverse of transconductance of the common diode-connected transistor. The proposed idea is by employing as larger transconductance of the common diode-connected transistor as possible. The effectiveness of the proposed idea has been proved through the HSPICE simulation for the current mode Max circuits with various numbers of input signals.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37C no.12
• /
• pp.1289-1295
• /
• 2012

An adaptive digital predistortion (PD) technique is proposed for linearization of power amplifiers (PAs) in multiple-input multiple-output (MIMO) transmitters. We consider a PD structure equipped with only one combined feedback path while conventional systems have multiple feedback paths. Hence, the proposed structure is much simpler than that of multiple feedback paths. Based on the structure, a new PD algorithm is derived. The simulation results show that linearization performance of the proposed method is almost the same as the conventional multiple feedback technique while the former is much simpler to implement than the latter.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.44 no.3 s.357
• /
• pp.65-71
• /
• 2007

We propose a pixel-level automatic calibration circuit scheme that initializes a capacitive fingerprint sensor LSI to eliminate the influence of the surface condition and environment, which is degraded by dirt during long-time use, process variation and ambient temperature. The sample chip is fabricated on $0.35{\mu}m$ standard CMOS process. The calibration is executed by optimizing the reference voltage in each pixel to make the sensor signals of all pixels the same. The calibration control circuit is composed of the sensing circuit and charge pumping circuit, and calibrates all pixels in a short time. 16-level gray scale fingerprint images can be captured to increase the accuracy of identification. This confirms that the scheme is effective for capturing consistent clear images during long-time use.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.21 no.5
• /
• pp.1325-1331
• /
• 1996

In this paper, a new current-memory circuit is proposed that reduces the clock feedthrough(CFT) error voltage causing total harmonic distortion(THD) increment in switched-current(SI) systems. Using PMOS transistor in CMOS complementary, the proposed one reduces output distortion current due to the CFT errorvoltage. A proposed current-memory is designed using a 1.2.mu.m CMOS process anda 1MHz sinusoidal signal having a 68.mu.A amplitude current is applied as input (sampling frequency:20MHz). It hasbeen shown from the simulation that the output distortion current effected by the CFT error voltage is reduced by approximately 10 times the error voltage of conventional one, THD is -57dB in case ofappling 1kHz frequency input signalwith 0.5 peak signal-to-bias current ratio.

• Journal of Korea Multimedia Society
• /
• v.13 no.3
• /
• pp.410-416
• /
• 2010

This paper presents the auto-extraction method that searches for the Mapping points in the calibration algorithm of the image acquired by the wide angle CCD camera. In this algorithm, we remove the noise from the distorted image and then obtain the edge image. Proposed method extracts the distortion point, comparing the threshold value of the histogram of the horizontal and vertical pixel lines in edge image. This processing step can be directly applied to the original image of the wide angle CCD camera output. Proposed method results are compared with hand-worked result image using the two wide angle CCD cameras having different angles with the difference value of the result images respectively. Experimental results show that proposed method can allocate the distortion-calibration constant of the wide angle CCD camera regardless of lens type, distortion shape and image type.

• Journal of Korea Multimedia Society
• /
• v.3 no.3
• /
• pp.263-271
• /
• 2000

In this paper, an acoustic echo canceller(AEC) is implemented by ADSP-2181. This AEC uses a noise robust adaptive algorithm and a postprocessing method which attenuates residual echo using cross-correlation between estimated error signal and microphone input signal. We propose new postprocessing method that uses two thresholds to prevent signal distortion after postprocessing and to improve the performance of AEC without extra computational burden. Through experiments using a 16 bit fixed-point DSP board (ADSP-2181 EZ-KIT Lite board), it is shown that the noise robust adaptive algorithm performs well in the double-talk situations and the convergence speed is comparable to NLMS. Using the postprocessor, ERLE is improved about 20 dB. As a result, the AEC with a postprocessor shows better performance than conventional ones.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.11C
• /
• pp.1079-1086
• /
• 2007

In this paper, we propose a CCP (Companding Combined with Predistorter) scheme that maintains BER (Bit Error Rate) performance of systems, while effectively reducing PAPR (Peak-to-Average Power Ratio) of OFDM (Orthogonal Frequency Division Multiplexing) signals, independent of nonlinear property of HPAs (High Power Amplifiers). The proposed CCP scheme combines the conventional companding scheme proposed by the authors with the predistorter, in order to prevent degradation of BER performance caused by signal distortion from nonlinear property of the HPA and the compression property of the companding scheme. Hence, the proposed scheme effectively reduces the PAPR and simultaneously maintains the BER performance, independent of nonlinear property of the HPA. Simulation results showed that the proposed CCP scheme provides the lowest total degradation for all the modulations, when comparing with the companding scheme proposed by Wang and the clipping scheme.