• Title/Summary/Keyword: Computation reduction

Search Result 443, Processing Time 0.024 seconds

Reduction of Dimension of HMM parameters in MLLR Framework for Speaker Adaptation (화자적응시스템을 위한 MLLR 알고리즘 연산량 감소)

  • Kim Ji Un;Jeong Jae Ho
    • Proceedings of the KSPS conference
    • /
    • 2003.05a
    • /
    • pp.123-126
    • /
    • 2003
  • We discuss how to reduce the number of inverse matrix and its dimensions requested in MLLR framework for speaker adaptation. To find a smaller set of variables with less redundancy, we employ PCA(principal component analysis) and ICA(independent component analysis) that would give as good a representation as possible. The amount of additional computation when PCA or ICA is applied is as small as it can be disregarded. The dimension of HMM parameters is reduced to about 1/3 ~ 2/7 dimensions of SI(speaker independent) model parameter with which speech recognition system represents word recognition rate as much as ordinary MLLR framework. If dimension of SI model parameter is n, the amount of computation of inverse matrix in MLLR is proportioned to O($n^4$). So, compared with ordinary MLLR, the amount of total computation requested in speaker adaptation is reduced to about 1/80~1/150.

  • PDF

Current Trend and Direction of Deep Learning Method to Railroad Defect Detection and Inspection

  • Han, Seokmin
    • International Journal of Internet, Broadcasting and Communication
    • /
    • v.14 no.3
    • /
    • pp.149-154
    • /
    • 2022
  • In recent years, the application of deep learning method to computer vision has shown to achieve great performances. Thus, many research projects have also applied deep learning technology to railroad defect detection. In this paper, we have reviewed the researches that applied computer vision based deep learning method to railroad defect detection and inspection, and have discussed the current trend and the direction of those researches. Many research projects were targeted to operate automatically without visual inspection of human and to work in real-time. Therefore, methods to speed up the computation were also investigated. The reduction of the number of learning parameters was considered important to improve computation efficiency. In addition to computation speed issue, the problem of annotation was also discussed in some research projects. To alleviate the problem of time consuming annotation, some kinds of automatic segmentation of the railroad defect or self-supervised methods have been suggested.

A New Approximate DCT Computation Based on Subband Decomposition and Its Application (서브밴드 분리에 근거한 새로운 근사 DCT 계산과 응용)

  • Jeong, Seong-Hwan
    • The Transactions of the Korea Information Processing Society
    • /
    • v.3 no.5
    • /
    • pp.1329-1336
    • /
    • 1996
  • In many image compression applications, the discrete cosine transform (DCY) is well known for is highly efficient coding performance. However, it produces undesirable block artifacts in low-bit rate coding. In addition, in many practical applications, faster computation and easier VLST implementation of DCT coefficients are also important issues. The removal of the block artifacts and faster DCT computation are therefor of practical interest. In this paper, a modified DCTcomputation scheme was investigated, which provides a simple efficient solution to the reduction of the block artifacts while achieving faster computation. We have applied the new ap-proach to the low-bit rate coding and decoding of images. Simulation results on real images have verified the improved performance of the proposed method over the standar d method.

  • PDF

On a Reduction of Pitch Search Time for IMBE Vocoder by Using the Spectral AMDF (SAMDF를 이용한 IMBE VOCODER의 피치 검색 시간 단축에 관한 연구)

  • 홍성훈
    • Proceedings of the Acoustical Society of Korea Conference
    • /
    • 1998.06c
    • /
    • pp.155-158
    • /
    • 1998
  • IMBE(Improved Multi-Band Excitation) vocoders exhibit good performance at low data rates. The major drawback to IMBE coders is their large computational requirements. In this paper, thus, we propose a new pitch search method that preserves the quality of the IMBE vocoder with reduced complexity. The basic idea is to reduce computation complexity of the pitch searching by using the SAMDF. Applying the proposed method to the IMBE vocoder, we can get approximately 52.02% searching time reduction in the pitch search. There is no difference in voice quality between conventional IMBE and proposed IMBE.

  • PDF

A Study on Computerization of the Sight Reduction (천측계산의 전산화에 관한 연구)

  • 윤여정
    • Journal of the Korean Institute of Navigation
    • /
    • v.12 no.1
    • /
    • pp.27-43
    • /
    • 1988
  • The tedious work, connected to the altitude correction, the computation of altitudes and aximuths and the plotting of the position lines, has been a objection to celestial position fixing method. But using a computer , the severe objection will be practically overruled. The author had already studied on computerization of the sight reduction partially. This paper is to confirm reliability of coordinate of the moon and the navigational planet calculated by computer programming and to suggest a method of calculating ship's position fixed by two position lines.

  • PDF

Analysis of Graphs Using the Signal Flow Matrix (신호 흐름 행렬에 의한 그래프 해석)

  • 김정덕;이만형
    • 전기의세계
    • /
    • v.22 no.4
    • /
    • pp.25-29
    • /
    • 1973
  • The computation of transmittances between arbitrary input and output nodes is of particular interest in the signal flow graph theory imput. The signal flow matrix [T] can be defined by [X]=-[T][X] where [X] and [Y] are input nose and output node matrices, respectively. In this paper, the followings are discussed; 1) Reduction of nodes by reforming the signal flow matrix., 2) Solution of input-output relationships by means of Gauss-Jordan reduction method, 3) Extension of the above method to the matrix signal flow graph.

  • PDF

(Prediction of reduction goals : deterministic approach) (리덕션 골의 예상: 결정적인 접근 방법)

  • 이경옥
    • Journal of KIISE:Software and Applications
    • /
    • v.30 no.5_6
    • /
    • pp.461-465
    • /
    • 2003
  • The technique of reduction goal prediction in LR parsing has several applications such as the computation of right context. An LR parser generating the set of pre-determined reduction goals was previously suggested. The set approach is nondeterministic, and so it is inappropriate in some applications. This paper suggests a deterministic technique to give a uniquely predictable reduction symbol.

Performance Analysis on DMB System with PAPR Reduction Techniques (PAPR 저감기법을 적용한 DMB 시스템의 성능분석)

  • 정영호;함영권;김환우
    • Journal of Broadcast Engineering
    • /
    • v.8 no.3
    • /
    • pp.238-249
    • /
    • 2003
  • Eureka 147 DAB system based on the orthogonal frequency division multiplexing (OFDM) was adopted as the transmission scheme for terrestrial digital multimedia broadcasting (DMB) system in Korea. The OFDM has several advantages comparing to the single carrier modulation. However, its high Peak-to-average power ratio (PAPR) increases the complexity of the D/A and A/D converters and reduces the efficiency of the high power amplifier To reduce the high PAPR of OFDM, various techniques such as clipping, peak windowing, companding, selected mapping (SLM), Partial transmit sequences (PTS), etc. have been proposed. In this paper, we propose modified configurations of SLM and PTS for effective implementation and evaluate the performance on the PAPR reduction of DMB system. The simulation results show that the modified SLM (MSLM) has merits in reducing the amount of computation and hardware complexity due to the reduction of the number of vector $P^{(U)}$, while satisfying the same performance and maintaining the same required bits (RB) for side information. With the same amount of computation and the same RB, the modified PTS (MPTS) is also shown to be better than PTS in the performance of PAPR reduction.

Shape Design of Frame Structures for Vibration Suppression and Weight Reduction

  • Hase, Miyahito;Ikeda, Masao
    • 제어로봇시스템학회:학술대회논문집
    • /
    • 2003.10a
    • /
    • pp.2246-2251
    • /
    • 2003
  • This paper proposes shape design of frame structures for vibration suppression and weight reduction. The $H_{\infty}$ norm of the transfer function from disturbance sources to the output points where vibration should be suppressed, is adopted as the performance index to represent the magnitude of vibration transfer. The design parameters are the node positions of the frame structure, on which constraints are imposed so that the structure achieves given tasks. For computation of Pareto optimal solutions to the two-objective design problem, a number of linear combinations of the $H_{\infty}$ norm and the total weight of the structure are considered and minimized. For minimization of the scalared objective function, a Lagrange function is defined by the objective function and the imposed constraints on the design parameters. The solution for which the Lagrange function satisfies the Karush-Kuhn-Tucker condition, is searched by the sequential quadratic programming (SQP) method. Numerical examples are presented to demonstrate the effectiveness of the proposed design method.

  • PDF