• Journal of Applied Reliability
• /
• v.9 no.2
• /
• pp.121-130
• /
• 2009

The power law process model the Rate of occurrence of failures(ROCOF) with monotonic trend during the operating time. However, the power law process is inappropriate when a non-monotonic trend in the failure data is observed. In this paper we deals with the reliability modeling of the failure process of large and complex repairable system whose rate of occurrence of failures shows the non-monotonic trend. We suggest a sectional model and a change-point test based on the Schwarz information criterion(SIC) to describe the non-monotonic trend. Maximum likelihood is also suggested to estimate parameters of sectional model. The suggested methods are applied to field data from an repairable system.

• Communications for Statistical Applications and Methods
• /
• v.10 no.3
• /
• pp.971-980
• /
• 2003

The tree method can be extended to multivariate responses, such as repeated measure and longitudinal data, by modifying the split function so as to accommodate multiple responses. Recently, some decision trees for multiple responses have been constructed by Segal (1992) and Zhang (1998). Segal suggested a tree can analyze continuous longitudinal response using Mahalanobis distance for within node homogeneity measures and Zhang suggested a tree can analyze multiple binary responses using generalized entropy criterion which is proportional to maximum likelihood of joint distribution of multiple binary responses. In this paper, we will modify CART procedure and suggest a new tree-based method that can analyze multiple binary responses using similarity measures.

• Proceedings of the IEEK Conference
• /
• 1998.06a
• /
• pp.933-935
• /
• 1998

The HMM-Net is an architecture for a neural network that implements a hidden markov model (HMM). The architecture is developed for the purpose of combining the discriminant power of neural networks with the time-domain modeling capability of HMMs. Criteria used for learning HMM-Net classifiers are maximum likelihood(ML) and minimization of mean squared error(MMSE). In this paper we propose an efficient learning method of HMM_Net classifiers using a ML-MMSE hybrid criterion and report the results of an experimental study comparing the performance of HMM_Net classifiers trained by the gradient descent algorithm with the above criteria. Experimental results for the isolated numeric digits from /0/ to /9/ show that the performance of the proposed method is better than the others in the repects of learning and recognition rates.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4E
• /
• pp.21-27
• /
• 1996

In this paper, we propose a discriminative training algorithm for the stochastic segment model (SSM) in continuous speech recognition. As the SSM is usually trained by maximum likelihood estimation (MLE), a discriminative training algorithm is required to improve the recognition performance. Since the SSM does not assume the conditional independence of observation sequence as is done in hidden Markov models (HMMs), the search space for decoding an unknown input utterance is increased considerably. To reduce the computational complexity and starch space amount in an iterative training algorithm for discriminative SSMs, a hybrid architecture of SSMs and HMMs is programming using HMMs. Given the segment boundaries, the parameters of the SSM are discriminatively trained by the minimum error classification criterion based on a generalized probabilistic descent (GPD) method. With the discriminative training of the SSM, the word error rate is reduced by 17% compared with the MLE-trained SSM in speaker-independent continuous speech recognition.

• International journal of advanced smart convergence
• /
• v.9 no.1
• /
• pp.98-108
• /
• 2020

We consider transmit antenna selection combined with power allocation for quadrature spatial modulation (QSM) systems to improve the error rate performance. The Euclidean distance-based joint optimization criterion is presented for transmit antenna selection and power allocation in QSM. It requires an exhaustive search and thus high computational complexity. Thus its reduced-complexity algorithm is proposed with a strategy of decoupling, which is employed to successively find transmit antennas and power allocation factors. First, transmit antennas are selected without considering power allocation. After selecting transmit antennas, power allocation factors are determined. Simulation results demonstrate considerable performance gains with lower complexity for transmit antenna selected QSM systems with power allocation, which can be achieved with limited rate feedback.

• 한국데이터정보과학회:학술대회논문집
• /
• 2002.06a
• /
• pp.155-167
• /
• 2002

In this paper we consider optimal designs of partially constant-stress life testing which is deviced for three-component mixed systems with the considerably long time. Mixed systems are jointed serial system with parallel system. Test items are run at both use condition and accelerated condition until a specified censoring time. The optimal criterion for the sample-proportion allocated to accelerated condition is to minimized asymptotic variance of the maximum likelihood estimators of the acceleration factor and hazard rates.

• 한국데이터정보과학회:학술대회논문집
• /
• 2006.04a
• /
• pp.163-169
• /
• 2006

This paper deals with the comparison of parameter estimation methods in a 3-parameter Kappa distribution which is sometimes used in flood frequency analysis. The method of moment estimation(MME), L-moment estimation(L-ME), and maximum likelihood estimation(MLE) are applied to estimate three parameters. The performance of these methods are compared by Monte-carlo simulations. Especially for computing MME and L-ME, ike dimensional nonlinear equations are simplied to one dimensional equation which is calculated by the Newton-Raphson iteration under constraint. Based on the criterion of the mean squared error, the L-ME is recommended to use for small sample size $(n\leq100)$ while MLE is good for large sample size.

• Communications for Statistical Applications and Methods
• /
• v.9 no.1
• /
• pp.87-100
• /
• 2002

Poorly-posed problems in the balanced discriminant analysis was considered. We restrict consideration to the case of observations and the number of variables are the same and small. When these problems exist, we do not use a maximum likelihood estimates(MLE) to estimate covariance matrices. Instead of MLE, an alternative estimate for the covariance matrices are proposed. This alternative method make good use of two regularization parameters, $\lambda$} and $\gamma$. A new test rule for the discriminant function is suggested and examined via limited hut informative simulation study. From the simulation study, it is shown that the suggested test rule gives better test result than other previously suggested method in terms of error rate criterion.

• Journal of the Korean Statistical Society
• /
• v.28 no.1
• /
• pp.73-92
• /
• 1999

In Bayesian model selection or testing problems, one cannot utilize standard or default noninformative priors, since these priors are typically improper and are defined only up to arbitrary constants. The resulting Bayes factors are not well defined. A recently proposed model selection criterion, the intrinsic Bayes factor overcomes such problems by using a part of the sample as a training sample to get a proper posterior and then use the posterior as the prior for the remaining observations to compute the Bayes factor. Surprisingly, such Bayes factor can also be computed directly from the full sample by some proper priors, namely intrinsic priors. The present paper explains how to derive intrinsic priors for simple tree ordered exponential means. Some numerical results are also provided to support theoretical results and compare with classical methods.

• Journal of Korean Society for Quality Management
• /
• v.17 no.1
• /
• pp.89-106
• /
• 1989

For Weibull distributed lifetimes, this paper presents asymptotically optimal accelerated life test plans for practical applications under intermittent inspection and type-I censoring. Computational results show that the asymptotic variance of a low quantile at the design stress as optimal criterion is insensitive to the number of inspections at overstress levels. Sensitivity analyses indicate that optimal plans are robust enough to moderate departures of estimated failure probabilities at the design and high stresses as input parameters to plan accelerated life tests from their true values. Monte Carlo simulation for small sample study on optimal accelerated life test plans developed by the asymptotic maximum likelihood theory is conducted. Simulation results suggest that optimal plans are satisfactory for sample size in practice.