• Title/Summary/Keyword: Recursive Method

Search Result 745, Processing Time 0.025 seconds

Evidence Extraction Method for Machine Reading Comprehension Model using Recursive Neural Network Decoder (디코더를 활용한 기계독해 모델의 근거 추출 방법)

  • Kyubeen Han;Youngjin Jang;Harksoo Kim
    • Annual Conference on Human and Language Technology
    • /
    • 2023.10a
    • /
    • pp.609-614
    • /
    • 2023
  • 최근 인공지능 시스템이 발전함에 따라 사람보다 높은 성능을 보이고 있다. 또한 전문 지식에 특화된 분야(질병 진단, 법률, 교육 등)에도 적용되고 있지만 이러한 전문 지식 분야는 정확한 판단이 중요하다. 이로 인해 인공지능 모델의 결정에 대한 근거나 해석의 중요성이 대두되었다. 이를 위해 설명 가능한 인공지능 연구인 XAI가 발전하게 되었다. 이에 착안해 본 논문에서는 기계독해 프레임워크에 순환 신경망 디코더를 활용하여 정답 뿐만 아니라 예측에 대한 근거를 추출하고자 한다. 실험 결과, 모델의 예측 답변이 근거 문장 내 등장하는지에 대한 실험과 분석을 수행하였다. 이를 통해 모델이 추론 과정에서 예측 근거 문장을 기반으로 정답을 추론한다는 것을 확인할 수 있었다.

  • PDF

Turbo MAP Decoding Algorithm based on Radix-4 Method (Radix-4 방식의 터보 MAP 복호 알고리즘)

  • 정지원;성진숙;김명섭;오덕길;고성찬
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.25 no.4A
    • /
    • pp.546-552
    • /
    • 2000
  • The decoding of Turbo-Code relies on the application of a soft input/soft output decoders which can be realized using maximum-a-posteriori(MAP) symbol estimator[l]. Radix-2 MAP decoder can not be used for high speed communications because of a large number of interleaver block size N. This paper proposed a new simple method for radix-4 MAP decoder based on radix-2 MAP decoder in order to reduce the interleave block size. A branch metrics, forward and backward recursive functions are proposed for applying to radix-4 MAP structure with symbol interleaver. Radix-4 MAP decoder shall be illustratively described and its error performance capability shall be compared to conventional radix-2 MAP decoder in AWGN channel.

  • PDF

Low-noise VLSI Implementation of Pipelined IIR Filters (파이프라인된 IIR 필터의 저잡음 VLSI구현)

  • 태기철;최정필;신승철;정진균
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.25 no.4B
    • /
    • pp.788-795
    • /
    • 2000
  • Scattered look-ahead pipelining method can be efficiently used for high sample rate or low-power applications of digital recursive filters. Although the pipelined filters are guaranteed to be stable by this method, these filters suffer from large round off noise when the poles are crowed within some critical regions. To avoid this problem, a low-noise implementation technique was proposed using constrained Remez exchange algorithm. By the constrained filter design approach, the desired filter spectrum is satisfied while some of the pole angles are constrained to avoid pole crowding within critical regions. In the proposed approach, to obtain improved spectrum characteristics or better round off noise properties, the radius of the angle-constrained pole is optimized depending on the direction of the pole movement.

  • PDF

An OFDM Receiver Scheme for Multipath Environment with Delay Profile over the Guard Interval (가드인터벌을 초월하는 지연프로필의 멀티패스채널에 적합한 OFDM수신시스템)

  • 주창복
    • Journal of the Institute of Convergence Signal Processing
    • /
    • v.4 no.4
    • /
    • pp.66-72
    • /
    • 2003
  • In the present paper some fundamentals about exceeding the guard interval of an OFDM system is going to discuss and will introduce method to cancel a ISI and ICI effects. In this paper, the performance characteristics of robust OFDM receiving system algorithm that is composed of a channel estimator using recursive least square(RLS) algorithm in time domain and of a ISI and ICI replica and subtracter using the channel coefficients are presented. By computer simulation method, the effect of suppression of ISI and ICI is demonstrated and compared the BER performance with a conventional 1-tap equalized system. Especially the presented scheme improves the BER performance at low SNRs more or so compared with the conventional one-tap equalized receiving configuration.

  • PDF

A Study on the Feature Extraction for High Speed Character Recognition -By Using Interative Extraction and Hierarchical Formation of Directional Information- (고속 문자 인식을 위한 특징량 추출에 관한 연구 - 방향정보의 반복적 추출과 특징량의 계층성을 이용하여 -)

  • 강선미;이기용;양윤모;양윤모;김덕진
    • Journal of the Korean Institute of Telematics and Electronics B
    • /
    • v.29B no.11
    • /
    • pp.102-110
    • /
    • 1992
  • In this paper, a new method of character recognition is proposed. It uses density information, in addition to positional and directional information generally used, to recognize a character. Four directional feature primitives are extracted from the thinning templates on the observation that the output of the templates have directional property in general. A simple and fast feature extraction scheme is possible. Features are organized from recursive nonary tree(N-tree) that corresponds to normalized character area. Each node of the N-tree has four directional features that are sum of the features of it's nine sub-nodes. Every feature primitive from the templates are added to the corresponding leaf and then summed to the upper nodes successively. Recognition can be accomplished by using appropriate feature level of N-tree. Also, effectiveness of each node's feature vector was tested by experiment. A method to implement the proposed feature vector organization algorithm into hardware is proposed as well. The third generation node, which is 4$\times$4, is used as a unit processing element to extract features, and it was implemented in hardware. As a result, we could observe that it is possible to extract feature vector for real-time processing.

  • PDF

On-line integration of structural identification/damage detection and structural reliability evaluation of stochastic building structures

  • Lei, Ying;Wang, Longfei;Lu, Lanxin;Xia, Dandan
    • Structural Engineering and Mechanics
    • /
    • v.63 no.6
    • /
    • pp.789-797
    • /
    • 2017
  • Recently, some integrated structural identification/damage detection and reliability evaluation of structures with uncertainties have been proposed. However, these techniques are applicable for off-line synthesis of structural identification and reliability evaluation. In this paper, based on the recursive formulation of the extended Kalman filter, an on-line integration of structural identification/damage detection and reliability evaluation of stochastic building structures is investigated. Structural limit state is expanded by the Taylor series in terms of uncertain variables to obtain the probability density function (PDF). Both structural component reliability with only one limit state function and system reliability with multi-limit state functions are studied. Then, it is extended to adopt the recent extended Kalman filter with unknown input (EKF-UI) proposed by the authors for on-line integration of structural identification/damage detection and structural reliability evaluation of stochastic building structures subject to unknown excitations. Numerical examples are used to demonstrate the proposed method. The evaluated results of structural component reliability and structural system reliability are compared with those by the Monte Carlo simulation to validate the performances of the proposed method.

Study of the Interaction between Tracked Vehicle and Terrain (궤도차량과 토양의 상호작용에 대한 연구)

  • Park, Cheon-Seo;Lee, Seung-Jong
    • Journal of the Korean Society for Precision Engineering
    • /
    • v.19 no.2
    • /
    • pp.140-150
    • /
    • 2002
  • The planar tracked vehicle model used in this investigation consists of two kinematically decoupled subsystems, i.e., the chassis subsystem and the track subsystem. The chassis subsystem includes the chassis frame, sprocket, idler and rollers, while the track subsystem is represented as a closed kinematic chain consisting of rigid links interconnected by revolute joints. In this study, the recursive kinematic and dynamic formulation of the tracked vehicle is used to find the vertical terce and the distance of an arbitrary track moved in the driving direction along the track. These distances and vertical forces obtained are used to get the deformation and sinkage of a terrain. The FEM(Finite Element Method) is adopted to analyze the interaction between tracked vehicle and terrain. The terrain is represented by a system of elements wish specified constitutive relationships and considered as a piecewise linear elastic, plastic and isotropic material. When the tracked vehicle is moving with different speeds on the terrain, the elastic and plastic deformations and the maximum sinkage for the four different types of isotropic soils are simulated.

A New Fast Algorithm for Short Range Force Calculation (근거리 힘 계산의 새로운 고속화 방법)

  • Lee, Sang-Hwan;Ahn, Cheol-O
    • 유체기계공업학회:학술대회논문집
    • /
    • 2006.08a
    • /
    • pp.383-386
    • /
    • 2006
  • In this study, we propose a new fast algorithm for calculating short range forces in molecular dynamics, This algorithm uses a new hierarchical tree data structure which has a high adaptiveness to the particle distribution. It can divide a parent cell into k daughter cells and the tree structure is independent of the coordinate system and particle distribution. We investigated the characteristics and the performance of the tree structure according to k. For parallel computation, we used orthogonal recursive bisection method for domain decomposition to distribute particles to each processor, and the numerical experiments were performed on a 32-node Linux cluster. We compared the performance of the oct-tree and developed new algorithm according to the particle distributions, problem sizes and the number of processors. The comparison was performed sing tree-independent method and the results are independent of computing platform, parallelization, or programming language. It was found that the new algorithm can reduce computing cost for a large problem which has a short search range compared to the computational domain. But there are only small differences in wall-clock time because the proposed algorithm requires much time to construct tree structure than the oct-tree and he performance gain is small compared to the time for single time step calculation.

  • PDF

Efficient Mixture IMM Algorithm for Speech Enhancement under Nonstationary Additive Colored Noise (시변가산유색잡음하의 음성 향상을 위한 효율적인 Mixture IMM 알고리즘)

  • 이기용;임재열
    • The Journal of the Acoustical Society of Korea
    • /
    • v.18 no.8
    • /
    • pp.42-47
    • /
    • 1999
  • In this paper, a mixture interacting multiple model (MIMM) algorithm is proposed to enhance speech contaminated by additive nonstationary noise. In this approach, a mixture hidden filter model (HFM) is used to model the clean speech and the noise process is modeled by a single hidden filter. The MIMM algorithm, however. needs large computation time because it is a recursive method based on multiple Kalman filters with mixture HFM. Thereby, a computationally efficient implementation of the algorithm is developed by exploiting the structure of the Kalman filtering equation. The simulation results show that the proposed method offers performance gain compared to the previous results in [4,5] with slightly increased complexity.

  • PDF

Sidelobe Reduction Method for Improvement of Airborne SAR Image (항공 SAR 영상 화질 개선을 위한 사이드로브 감소 기법)

  • Shin, Hee-Sub;Ok, Jae-Woo;Woo, Jae-Choon
    • The Journal of Korean Institute of Electromagnetic Engineering and Science
    • /
    • v.26 no.11
    • /
    • pp.1027-1030
    • /
    • 2015
  • In the airborne SAR, the motion errors induced by atmospheric turbulence decrease the resolution and increase the sidelobes. If the sidelobes are not properly compensated, the image quality is degraded. Thus, in this paper, we have introduced the sidelobe reduction method to increase the image quality. After we calculate the scene center based on the estimated squint angle for the flight path partitioned by the subaperture technique, we perform the motion compensation for the scene center. Then, after we perform the recursive sidelobe reduction for the region of interest in the reconstructed SAR image, we extend it for the full image.