• The Journal of the Acoustical Society of Korea
• /
• v.13 no.1E
• /
• pp.64-70
• /
• 1994

Predictive neural network models are powerful speech recognition models based on a nonlinear pattern prediction. But those models suffer from poor discrimination between acoustically similar words. In this paper we propose an discriminative training algorithm for predictive neural network models. This algorithm is derived from GPD (Generalized Probabilistic Descent) algorithm coupled with MCEF(Minimum Classification Error Formulation). It allows direct minimization of a recognition error rate. Evaluation of our training algoritym on ten Korean digits shows its effectiveness by 30% reduction of recognition error.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1994.06a
• /
• pp.997-1002
• /
• 1994

Predictive neural network models are powerful speech recognition models based on a nonlinear pattern prediction. Those models can effectively normalize the temporal and spatial variability of speech signals. But those models suffer from poor discrimination between acoustically similar words. In this paper, we propose a discriminative training algorithm for predictive neural network models based on a generalized probabilistic descent (GPD) algorithm and minimum classification error formulation (MCEF). The Evaluation of our training algorithm on ten Korean digits shows its effectiveness by 40% reduction of recognition error.

• The KIPS Transactions:PartB
• /
• v.15B no.1
• /
• pp.53-60
• /
• 2008

This paper proposes a new multiobjective optimization method for discriminative training of hidden Markov models (HMMs) used as the recognizer for automatic lipreading. While the conventional Baum-Welch algorithm for training HMMs aims at maximizing the probability of the data of a class from the corresponding HMM, we define a new training criterion composed of two minimization objectives and develop a global optimization method of the criterion based on simulated annealing. The result of a speaker-dependent recognition experiment shows that the proposed method improves performance by the relative error reduction rate of about 8% in comparison to the Baum-Welch algorithm.

• Journal of Korea Multimedia Society
• /
• v.24 no.1
• /
• pp.13-20
• /
• 2021

A Wasserstein GAN(WGAN), optimum in terms of minimizing Wasserstein distance, still suffers from inconsistent convergence or unexpected output due to inherent learning instability. It is widely known some kinds of restriction on the discriminative function should be considered to solve such problems, which implies the importance of Lipschitz continuity. Unfortunately, there are few known methods to satisfactorily maintain the Lipschitz continuity of the discriminative function. In this paper we propose techniques to stably maintain the Lipschitz continuity of the discriminative function by adding effective regularization terms to the objective function, which limit the magnitude of the gradient vectors of the discriminator to one or less. Extensive experiments are conducted to evaluate the performance of the proposed techniques, which shows the single-sided penalty improves convergence compared with the gradient penalty at the early learning process, while the proposed additional penalty increases inception scores by 0.18 after 100,000 number of learning.

• The Journal of the Acoustical Society of Korea
• /
• v.27 no.8
• /
• pp.443-449
• /
• 2008

One of the key issues in practical speech processing is to achieve robust Voice Activity Deteciton (VAD) against the background noise. Most of the statistical model-based approaches have tried to employ equally weighted likelihood ratios (LRs), which, however, deviates from the real observation. Furthermore voice activities in the adjacent frames have strong correlation. In other words, the current frame is highly correlated with previous frame. In this paper, we propose the effective VAD approach based on a minimum classification error (MCE) method which is different from the previous works in that different weights are assigned to both the likelihood ratio on the current frame and the decision statistics of the previous frame.

• Culinary science and hospitality research
• /
• v.6 no.3
• /
• pp.167-192
• /
• 2000

The purpose of this study is to research consumer's action style. and then to make a marketing strategies. The strategic plan to consumer's action style on the fast-food industry were as follows; First, the group that coefficient of utilization is so frequent and an age is young are requested concentrative or discriminative marketing as the price discrimination and market segementation, the price value of the products value has to manage effectively, Second, establishment of corporate image and improvement of products image are requested. The result of this study, fast-food industry will be needed marketing activities that are discriminative strategies, positioning strategies, education training, and customer management. And system construction that is equivalent to customer needs, and the continuous coustomer studies are requested.

• The Journal of the Acoustical Society of Korea
• /
• v.19 no.2
• /
• pp.83-89
• /
• 2000

In this paper, we proposed a discriminative adaptation method for PMC algorithm and achieved improved speech recognition rate. For the adaptation, we adopted modified PMC(MPMC) which is a variant of PMC and discriminatively adapted the association factor for each mixture of the HMM in the MPMC. From the recognition experiments, the proposed method showed better recognition rate than the conventional PMC. Also, compared with STAR algorithm which is another model parameter compensation method, the proposed method showed superior performance when the SNR is very low and the adaptation data is not sufficient.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.3
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.

• Journal of information and communication convergence engineering
• /
• v.21 no.2
• /
• pp.139-144
• /
• 2023

Engineers prefer deep neural networks (DNNs) for solving computer vision problems. However, DNNs pose two major problems. First, neural networks require large amounts of well-labeled data for training. Second, the covariate shift problem is common in computer vision problems. Domain adaptation has been proposed to mitigate this problem. Recent work on adversarial-learning-based unsupervised domain adaptation (UDA) has explained transferability and enabled the model to learn robust features. Despite this advantage, current methods do not guarantee the distinguishability of the latent space unless they consider class-aware information of the target domain. Furthermore, source and target examples alone cannot efficiently extract domain-invariant features from the encoded spaces. To alleviate the problems of existing UDA methods, we propose the mixup regularization in adversarial discriminative domain adaptation (ADDA) method. We validated the effectiveness and generality of the proposed method by performing experiments under three adaptation scenarios: MNIST to USPS, SVHN to MNIST, and MNIST to MNIST-M.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1993.06a
• /
• pp.242-246
• /
• 1993

예측신경회로망 모델은 다층 퍼셉트론을 연속되는 음성특징 벡터간의 비선형예측기로 사용하는 동적인 음성인식 모델이다. 이 모델은 음성의 동적인 특성을 인식에 이용하고 연속음성인식으로의 확장이 용이한 우수한 인식 모델이다. 그러나, 예측신경회로망 모델은 음운학적으로 유사한 음성구간에서의 변별력이 낮다는 문제점이 있다. 그것은 기존의 학습 알고리즘이 다른 어휘와의 거리는 고려하지 않고 대상어휘의 예측오차만 최소화시키기 때문이다. 따라서, 본 논문에서는 직접 인식오차를 최소화시키는 GPD알고리즘에 의해 유사어휘간의 거리를 고려하는 변별력있는 학습 알고리즘을 제안한다.