• Journal of Korea Multimedia Society
• /
• v.4 no.5
• /
• pp.414-422
• /
• 2001

This paper proposes a classifier of music genre using time delay neural network(TDNN) fur an audio data retrieval systems. The classifier considers eight kinds of genres such as Blues, Country, Hard Core, Hard Rock, Jazz, R&B(Soul), Techno and Trash Metal. The comparative unit to classify the genres is a melody between bars. The melody pattern is extracted based un snare drum sound which represents the periodicity of rhythm effectively. The classifier is constructed with the TDNN and uses fourier transformed feature vector of the melody as input pattern. We experimented the classifier on eighty training data from ten musics for each genres and forty test data from five musics for each genres, and obtained correct classification rates of 92.5% and 60%, respectively.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.21 no.2
• /
• pp.534-540
• /
• 1996

This paper presents an Adaptive Time-Delay Recurrent Neural Network (ATRN) for learning and recognition of temporal correlations of temporal patterns. The ATRN employs adaptive time-delays and recurrent connections, which are inspired from neurobiology. In the ATRN, the adaptive time-delays make the ATRN choose the optimal values of time-delays for the temporal location of the important information in the input parrerns, and the recurrent connections enable the network to encode and integrate temporal information of sequences which have arbitrary interval time and arbitrary length of temporal context. The ATRN described in this paper, ATNN proposed by Lin, and TDNN introduced by Waibel were simulated and applied to the chaotic time series preditcion of Mackey-Glass delay-differential equation. The simulation results show that the normalized mean square error (NMSE) of ATRN is 0.0026, while the NMSE values of ATNN and TDNN are 0.014, 0.0117, respectively, and in temporal learning, employing recurrent links in the network is more effective than putting multiple time-delays into the neurons. The best performance is attained bythe ATRN. This ATRN will be sell applicable for temporally continuous domains, such as speech recognition, moving object recognition, motor control, and time-series prediction.

• 제어로봇시스템학회:학술대회논문집
• /
• 1999.10a
• /
• pp.214-217
• /
• 1999

In this paper, genetic algorithm (GA) is implemented to search for the optimal structures (i.e. the kind of neural networks, the number of inputs and hidden neurons) of neural networks which are used approximating a given nonlinear function. Two kinds of neural networks, i.e. the multilayer feedforward [1] and time delay neural networks (TDNN) [2] are involved in this paper. The synapse weights of each neural network in each generation are obtained by associated training algorithms. The simulation results of nonlinear function approximation are given out and some improvements in the future are outlined.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.4
• /
• pp.46-63
• /
• 1995

This thesis described an unlimited vocabulary connected speech recognition system using Time Delay Neural Network(TDNN). The recognition unit is the diphone unit which includes the transition section of two phonemes, and the number of diphone unit is 329. The recognition processing of korean connected speech is composed by three part; the feature extraction section of the input speech signal, the diphone recognition processing and post-processing. In the feature extraction section, the extraction of diphone interval in input speech signal is carried and then the feature vectors of 16th filter-bank coefficients are calculated for each frame in the diphone interval. The diphone recognition processing is comprised by the three stage hierachical structure and is carried using 30 Time Delay Neural Networks. particularly, the structure of TDNN is changed so as to increase the recognition rate. The post-processing section, mis-recognized diphone strings are corrected using the probability of phoneme transition and the probability o phoneme confusion and then the eojeols (Korean word or phrase) are formed by combining the recognized diphones.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.483-489
• /
• 2020

This paper proposes acoustic model training using self-attention for low-resource speech recognition. In low-resource speech recognition, it is difficult for acoustic model to distinguish certain phones. For example, plosive /d/ and /t/, plosive /g/ and /k/ and affricate /z/ and /ch/. In acoustic model training, the self-attention generates attention weights from the deep neural network model. In this study, these weights handle the similar pronunciation error for low-resource speech recognition. When the proposed method was applied to Time Delay Neural Network-Output gate Projected Gated Recurrent Unit (TNDD-OPGRU)-based acoustic model, the proposed model showed a 5.98 % word error rate. It shows absolute improvement of 0.74 % compared with TDNN-OPGRU model.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.3
• /
• pp.334-341
• /
• 2003

In this paper, we proposed a recurrent time delayed neural network(RTDNN) controller which compensate a output of neural network controller. Even if learn by neural network controller, it can occur an bad results from disturbance or load variations. So in order to adjust above case, we used the fuzzy compensator to get an expected results. And the weight of main neural network can be changed with the result of learning a inverse model neural network of plant, so a expected dynamic characteristics of plant can be got. As the results of simulation through the second order plant, we confirmed that the proposed recurrent time delayed neural network controller get a good response compare with a time delayed neural network(TDU) controller. We implemented the controller using the DSP processor and applied in a hydraulic servo system. And then we observed an experimental results.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2010.10a
• /
• pp.108-109
• /
• 2010

본 연구에서는 비선형예측기법으로서 그 우수성을 인정받고 있는 인공신경망모형을 사용하여 컨테이너 물동량 예측을 수행하였다. 그러나 인공신경망모형을 사용해 시계열의 예측결과를 ARIMA모형과 같이 널리 알려진 다른 전통적인 수요예측기법들과 비교 평가한 과거 연구들을 보게 되면 각기 주장하는 바와 그 결론이 상반됨을 알 수 있다. 그래서 인공신경망의 예측성과를 높이기 위한 기존의 선행연구들의 다양한 시도들을 바탕으로 국내 항만의 컨테이너물동량을 예측하고, 그를 통해 여러 모형간의 비교 검증작업을 수행하였다.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.4
• /
• pp.419-426
• /
• 1999

A novel motion analysis system is presented in this paper. The proposed system is inspired by processing functions observed in the fly visual system, which detects changes in input light intensities, determines motion on both the local and the wide-field levels. The system has several differences from conventional motion analysis system. First, conventional systems usually focused on matching similar feature or optical flow, but neural network is applied in this system. Back propagation is used by learning method, and Tine Delay Neural Network (TDNN) is also used as analysis method. Second, while conventional systems usually limited on only two frames of sequence, the proposed system accept multiple frames of sequence. The experimental results showed a 94.7% correct rate with a speed of 71.47 milli seconds for real and synthetic images.

• Journal of Industrial Technology
• /
• v.21 no.B
• /
• pp.91-98
• /
• 2001

In this paper, center detection and motion analysis of a moving object are studied. Kohonen's self-organizing neural network models are used for the moving objects tracking and time delay neural networks are used for dynamic characteristic analysis. Instead of objects brightness, neuron projections by Kohonen Networks are used. The motion of target objects can be analyzed by using the differential neuron image between the two projections. The differential neuron image which is made by two consecutive neuron projections is used for center detection and moving objects tracking. The two differential neuron images which are made by three consecutive neuron projections are used for the moving trajectory estimation. It is possible to distinguish 8 directions of a moving trajectory with two frames and 16 directions with three frames.

• Journal of Intelligence and Information Systems
• /
• v.5 no.1
• /
• pp.49-64
• /
• 1999

본 연구는 백화점 고객이 신용 카드 신청 요구 시에 작성되는 가입 정보 및 사용되고 있는 고객의 거래 정보는 카드 사용 패턴으로 신용도를 예측하는 여러 방법론을 제시하고 성능을 비교하였다. 가입 정보를 분석하기 위해 역전파 신경망(Back-Propagation Neural Network, BPNN), 사례기반추론(Case-Based reasoning)을, 거래 정보를 분석하기 위해 역전파 신경망과 더불어 시간지연 신경망(Time-Delayed Neural Network, TDNN)을 각각 사용하여 그 결과를 비교하였다. 또한 전체시스템의 적중률을 높이기 위햐여, ID3와 신경망을 이용한 Meta-Leaning 방법을 제시하였으며, Meta-Learning 방법과 다른 방법들을 비교, 분석을 하였다. 본 연구에서는 모형 수립과 검증을 위하여 T백화점의 실제 신용 카드 가입 고객 데이터를 이용하여 실험하였다. 데이터의 성격에 따라 각 모델의 예측력에는 차이가 나타났으나, 신경망 모형의 예측력이 우수하였으며, 시간적 특성을 고려하는 시간지연 신경회로망 모형의 예측력은 더욱 우수하게 나타났다. 또한 Meta-Learning 모형을 사용하면 예측력이 더 높아진다는 것을 확인할 수 있었다.