• ETRI Journal
• /
• 제27권4호
• /
• pp.461-464
• /
• 2005

Recently, a variety of confidence measures for utterance verification has been studied to improve speech recognition performance by rejecting out-of-vocabulary inputs. Most of the conventional confidence measures for utterance verification are based primarily on hypothesis testing or an approximated posterior probability, and their performances depend on the robustness of an alternative hypothesis or the prior probability. We introduce a novel confidence measure called a search confusion rate (SCR), which does not require an alternative hypothesis or the approximation of posterior probability. Our confusion-based approach shows better performance in additive noise-corrupted speech as well as in clean speech.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2007년도 하계종합학술대회 논문집
• /
• pp.253-254
• /
• 2007

We investigate error patterns in compressed multi-view videos and propose a multi-hypothesis algorithm, which is adaptive to the error patterns. Different from multi-hypothesis algorithms for mono-view sequences, the proposed algorithm exploits inter-view correlations in multi-view sequences as well as spatio-temporal correlations. Simulation results demonstrate that the proposed algorithm effectively protects the quality of reconstructed videos against transmission errors.

• Industrial Engineering and Management Systems
• /
• 제11권4호
• /
• pp.385-389
• /
• 2012

With the development of computer storage and the rapidly growing ability to process large amounts of data, the multivariate control charts have received an increasing attention. The existing univariate and multivariate control charts are a single hypothesis testing approach to process mean or variance by using a single statistic plot. This paper proposes a multiple hypothesis approach to developing a new multivariate control scheme. Plotted Hotelling's $T^2$ statistics are used for computing the corresponding p-values and the procedure for controlling the false discovery rate in multiple hypothesis testing is applied to the proposed control scheme. Some numerical simulations were carried out to compare the performance of the proposed control scheme with the ordinary multivariate Shewhart chart in terms of the average run length. The results show that the proposed control scheme outperforms the existing multivariate Shewhart chart for all mean shifts.

• 대한전기학회논문지
• /
• 제45권1호
• /
• pp.139-145
• /
• 1996

The input estimation method estimates maneuvering input acceleration in order to track a maneuvering target. In this paper, the optimal input estimator is derived by choosing the MAP hypothesis among maneuvering input transition hypotheses under the assumption that a maneuvering input acceleration is a semi-Markov process. The optimal input estimation method cannot be realized because the optimal filter should consider every maneuver onset time hypothesis from filter starting time to current time which increase rapidly. Hence the suboptimal filter using a sliding window is proposed. Since the proposed method can consider all hypotheses of input transitions inside the window, it is general enough to include Bogler's input estimation method. Simulation results show, however, that we can obtain a good performance even when the filter considering just one input transition in the window is used. (author). 9 refs., 3 figs., 1 tab.

• 한국인쇄학회지
• /
• 제30권2호
• /
• pp.69-78
• /
• 2012

This research attempts to analyze how attributes of ad model influence on brand familiarity, personality, and preference among consumers. Attributes of ad model was classified into three; attractiveness, professionalism, and reliability. One hundred of consumers who lived in Daejeun City were selected as respondents for this research. The result of hypothesis I showed that attributes of ad model significantly influenced on brand familiarity. it was the result of hypothesis II that both Attractiveness and professionalism of ad model affected brand personality. It was also found that attractiveness and professionalism are significantly brand preference from the result of hypothesis III. In conclusion, findings showed that attractiveness of ad model mostly influenced on brand familiarity, personality, preference.

• Communications for Statistical Applications and Methods
• /
• 제3권2호
• /
• pp.69-76
• /
• 1996

In a two-way contingency table, the analyst is most interested in the hypotheses of either homogeneity or independence. For testing this as a null hypothesis, Pearson's ${\chi}^2$ statistic is most commonly used in practice. Once the null Hypothesis is rejected, he will further search forcells which caused the rejection of the null hypothesis. For this purpose, so called cell${\chi}^2$ components are used. In this paper, we derive the influence function of an obsevation to the ${\chi}^2$ statistic, with which cells with high influence can be identified.

• Communications for Statistical Applications and Methods
• /
• 제11권3호
• /
• pp.643-655
• /
• 2004

L$_1$-estimator in the linear regression model is widely recognized to have superior robustness in the presence of vertical outliers. While the L$_1$-estimation procedures and algorithms have been developed quite well, less progress has been made with the hypothesis test in the multiple L$_1$-regression. This article suggests computer-intensive resampling approaches, jackknife and bootstrap methods, to estimating the variance of L$_1$-estimator and the scale parameter that are required to compute the test statistics. Monte Carlo simulation studies are performed to measure the power of tests in small samples. The simulation results indicate that bootstrap estimation method is the most powerful one when it is employed to the likelihood ratio test.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제10권10호
• /
• pp.5095-5111
• /
• 2016

The Probability Hypothesis Density (PHD) filter is a suboptimal approximation and tractable alternative to the multi-target Bayesian filter based on random finite sets. However, the PHD filter fails to track newborn targets when the target birth intensity is unknown prior to tracking. In this paper, a dual detection-guided newborn target intensity PHD algorithm is developed to solve the problem, where two schemes, namely, a newborn target intensity estimation scheme and improved measurement-driven scheme, are proposed. First, the newborn target intensity estimation scheme, consisting of the Dirichlet distribution with the negative exponent parameter and target velocity feature, is used to recursively estimate the target birth intensity. Then, an improved measurement-driven scheme is introduced to reduce the errors of the estimated number of targets and computational load. Simulation results demonstrate that the proposed algorithm can achieve good performance in terms of target states, target number and computational load when the newborn target intensity is not predefined in multi-target tracking systems.

• ETRI Journal
• /
• 제43권3호
• /
• pp.511-523
• /
• 2021

The World Health Organization provides guidelines for managing the particulate matter (PM) level because a higher PM level represents a threat to human health. To manage the PM level, a procedure for measuring the PM value is first needed. We use a PM sensor that collects the PM level by laser-based light scattering (LLS) method because it is more cost effective than a beta attenuation monitor-based sensor or tapered element oscillating microbalance-based sensor. However, an LLS-based sensor has a higher probability of malfunctioning than the higher cost sensors. In this paper, we regard the overall malfunctioning, including strange value collection or missing collection data as anomalies, and we aim to detect anomalies for the maintenance of PM measuring sensors. We propose a novel architecture for solving the above aim that we call the hypothesis pruning generative adversarial network (HP-GAN). Through comparative experiments, we achieve AUROC and AUPRC values of 0.948 and 0.967, respectively, in the detection of anomalies in LLS-based PM measuring sensors. We conclude that our HP-GAN is a cutting-edge model for anomaly detection.

• Journal of Acupuncture Research
• /
• 제36권1호
• /
• pp.21-27
• /
• 2019

Background: The intelligent tissue hypothesis holds that an organ's states (be it normal function or stressed states) are reflected in real time at its related acupuncture points (acupoints), causing physical, real-time changes in the local tissue. The experiment was devised to test this. Methods: The patient drank chilled water while the impedance at 6 stomach-related acupoints was monitored in real time (sample rate of 1 kHz). Any changes in the local tissue at the acupoints ought to be reflected in changes in the impedance. Results: The impedance at every test acupoint showed a response to the chilled water being ingested. Also, the duodenal pacesetter and the stomach's slow waves were clearly visible in the impedance pattern at all the acupoints. Conclusion: Hence, many separate details of the stomach's function were reflected at these acupoints. The duodenal features were consistent with the traditional indications for these acupoints, which are noted to be able to treat intestinal conditions. Therefore, the results were consistent with the hypothesis and also provided a possible explanation for how the use of these acupoints is able to treat intestinal conditions.