• ETRI Journal
• /
• v.30 no.5
• /
• pp.750-752
• /
• 2008

This paper provides a novel method of detecting key captions containing player information in golf videos. We use the color pattern of captions and its repetition property to determine the key captions. The experimental results show that the proposed method achieves a much higher accuracy than existing methods.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.241-243
• /
• 2020

The high computational complexity of deep learning algorithms has led to the development of specialized hardware architectures. However, soft errors (bit flip) may occur in these hardware systems due to voltage variation and high energy particles. Many error correction methods have been proposed to counter this problem. In this work, we analyze an error correction mechanism based on repetition codes and an activation function. We test this method by injecting errors into weight filters and define an ideal error rate range in which the proposed method complements the accuracy of the model in the presence of error.

• Korean Journal of Cognitive Science
• /
• v.24 no.4
• /
• pp.365-391
• /
• 2013

Negative repetition effect (NRE) refers to the phenomenon that the accuracy of report is impaired when a target was flanked by the same distractor than when by alternative distractor. To probe the nature of NRE, this study introduced attention window(s) indicating the positions where a target might be presented, and non-attention window(s) where a target could not be presented. Attention windows are supposed to help participants detect targets readily. Two among three positions are indicated by attention windows in Exp. 1, and a single large attention window encompassing central two positions among four positions was used in Exp. 2, and either large or small attention window was used depending on the number of target candidates in Exp. 3. In the result of three experiments, NREs were consistently observed when both positions of a target and a distractor were indicated by previous attention windows. However, NREs (including its tendency) and its opposite, PREs were observed when a distractor was presented in the non-attention position, depending on its distance from the target and the size of attention window. It seems that this pattern of repetition effects is hard to be explained by repetition blindness hypothesis (Kanwisher, 1991), positional uncertainty hypothesis (Keren & Boer, 1985), and inhibitory attention capture hypothesis (Kwak et al., 1993). Instead it was proposed that shifting of spatial attention should be considered accordingly with the structure of stimulus display. The promising role of this task was discussed in studying the relation of attention and perception.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.11
• /
• pp.1634-1640
• /
• 2017

The study aims to distinguish hemiplegic gait and normal gait using simple wearable device and classification algorithm. Thus, we developed a wearable system equipped three axis accelerometer and three axis gyroscope. The developed wearable system was verified by clinical experiment. In experiment, twenty one normal subjects and twenty one patients undergoing stroke treatment were participated. Based on the measured inertial signal, a random forest algorithm was used to classify hemiplegic gait. Four-fold cross validation was applied to ensure the reliability of the results. To select optimal attributes, we applied the forward search algorithm with 10 times of repetition, then selected five most frequently attributes were chosen as a final attribute. The results of this study showed that 95.2% of accuracy in hemiplegic gait and normal gait classification and 77.4% of accuracy in hemiplegic-side and normal gait classification.

• Korean Journal of Cognitive Science
• /
• v.33 no.1
• /
• pp.1-22
• /
• 2022

Lavie's perceptual load theory (Lavie, 1995) proposes that the influence of distractors would be blocked as the load gets higher. Studies of perceptual load have usually adopted the flanker task, developed by Eriksen and Eriksen (1974), which measures reaction time on the target flanked by distractors. In the post-cueing forced task, participants should report the identity of the target cued later, and negative repetition effect (NRE) has often been observed. NRE means the effect that the accuracy of identification is worse when the target is flanked by the same nontargets than when flanked by different nontargets. This study has tried to check whether perceptual load has an effect on identification rate and NRE. Experiment 1 manipulated the similarity between targets and a distractor, and observed a tendency of NRE, but not the effect of perceptual load. Experiment 2 used 4, 2 (in two kinds of diagonal arrangement), or none distractors of the same identity to burden more perceptual load. NRE was significant and perceptual load showed significance but not a linear trend. Experiment 3 checked again whether NRE would be varied according to two levels of perceptual load strengthened by positional variability of load stimuli, but did not find the effect of perceptual load. It is concluded that perceptual load might have a limited effect on the early stage of perceptual processing due to divided attentional processing of the targets briefly exposed. Implications of this study were discussed.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.44 no.2
• /
• pp.58-65
• /
• 2021

Meta-heuristic algorithms have been developed to efficiently solve difficult problems and obtain a global optimal solution. A common feature mimics phenomenon occurring in nature and reliably improves the solution through repetition. And at the same time, the probability is used to deviate from the regional optimal solution and approach the global optimal solution. This study compares the algorithm created based on the above common points with existed SA and HS to show advantages in time and accuracy of results. Existing algorithms have problems of low accuracy, high memory, long runtime, and ignorance. In a two-variable polynomial, the existing algorithms show that the memory increases and the accuracy decrease. In order to improve the accuracy, the new algorithm increases the number of initial inputs and increases the efficiency of the search by introducing a direction using vectors. And, in order to solve the optimization problem, the results of the last experiment were learned to show the learning effect in the next experiment. The new algorithm found a solution in a short time under the experimental conditions of long iteration counts using a two-variable polynomial and showed high accuracy. And, it shows that the learning effect is effective in repeated experiments.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.3
• /
• pp.34-41
• /
• 2014

Through applying hostile radar signals that use stagger PRI to PRI transform in real time, we can analyze stagger PRI and calculate the future TOA for active cancellation by using measured TOA and estimated PRI. We shows the effect of the errors that are contained in PRI and measured TOA. Then, it will suggest the technique to improve the accuracy of estimated PRI and the TOA averaging method for reducing the effect of measured TOA error. Finally, we will show that accuracy of estimated future TOA that is calculated by proposed scheme is higher than that of future TOA that is simply calculated with TDOA and newest TOA through comparing RMSE performance.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.9
• /
• pp.4264-4279
• /
• 2017

In this paper, maximum likelihood-based automatic lexicon generation using mixed-syllables is proposed for unlimited vocabulary voice interface for East Asian languages (e.g. Korean, Chinese and Japanese) in AI-assistant based interaction with mobile devices. The conventional lexicon has two inevitable problems: 1) a tedious repetition of out-of-lexicon unit additions to the lexicon, and 2) the propagation of errors during a morpheme analysis and space segmentation. The proposed method provides an automatic framework to solve the above problems. The proposed method produces a level of overall accuracy similar to one of previous methods in the presence of one out-of-lexicon word in a sentence, but the proposed method provides superior results with the absolute improvements of 1.62%, 5.58%, and 10.09% in terms of word accuracy when the number of out-of-lexicon words in a sentence was two, three and four, respectively.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2023.05a
• /
• pp.345-346
• /
• 2023

As the shape and size of detectable objects diversifying recognition and segmentation algorithms have been developed to acquire accurate shape information. Although a high density of data captured by the repetition of scanning improves the accuracy of algorithms the high dense data decreases the efficiency due to its large size. This paper proposes standardization algorithms using the feature of structural members on indoor point cloud data to improve the process. First of all we determine the reduction rate of the density based on the features of the target objects then the data reduction algorithm compresses the data based on the reduction rate. Second the data arrangement algorithm rotates the data until the normal vector of data is aligned along the coordinate axis to allow the following algorithms to operate properly. Final the data arrangement algorithm separates the rotated data into their leaning axis. This allows reverse engineering of indoor point clouds to obtain the efficiency and accuracy of refinement processes.

• Korean Journal of Cognitive Science
• /
• v.15 no.2
• /
• pp.17-22
• /
• 2004

Two experiments were executed to investigate the effect of uniform connectedness systematically using the identification task of briefly exposed forms. Previous study observed negative repetition effected (i.e., NRE) in the identification task of two parentheses either connected or disconnected vertically, which was interpreted as an evidence against the hypothesis of uniform connectedness. Experiment I tested the hypothesis that NRE resulted from the Perceptual set or anticipating disconnected displays. Experiment 2 tested the hypothesis that NRE resulted from relatively shorter exposure time. Using partial report task asking participants to report only the cued target and whole report task asking them to report the whole pattern with only connected displays, experiment 1 observed NRE respectively. Experiment 2, with longer exposure time equivalent to 83% accuracy and response bias controlled by use of catch trials, obtained the same NRE. Those results seems to indicate that uniformly connected forms were processed analytically by perceivers without task demand and futhermore, the hypothesis of uniform connectedness as a principle of perceptual organization is not plausible.