• Journal of the Korea Society of Computer and Information
• /
• v.21 no.4
• /
• pp.39-46
• /
• 2016

While the difficulty of the music can be classified by a variety of standard, conventional methods are classified by the subjective judgment based on the experience of many musicians or conductors. Music score is difficult to evaluate as there is no quantitative criterion to determine the degree of difficulty. In this paper, we propose a new classification method for determining the degree of difficulty of the music. In order to determine the degree of difficulty, we convert the score, which is expressed as a traditional music score, into electronic music sheet. Moreover, we calculate information about the elements needed to play sheet music by distance of notes, tempo, and quantifying the ease of interpretation. Calculating a degree of difficulty of the entire music via the numerical data, we suggest the difficulty evaluation of the score, and show the difficulty of music through experiments.

• International Journal of Internet, Broadcasting and Communication
• /
• v.15 no.3
• /
• pp.66-72
• /
• 2023

Little attention appears to have been paid to the relevance of learning a good representation function in solving long tail tasks. Therefore, we propose a new loss function to ensure a good representation is learnt while learning to classify. We call this loss function Triplet Class-Wise Difficulty-Based (TriCDB-CE) Loss. It is a combination of the Triplet Loss and Class-wise Difficulty-Based Cross-Entropy (CDB-CE) Loss. We prove its effectiveness empirically by performing experiments on three benchmark datasets. We find improvement in accuracy after comparing with some baseline methods. For instance, in the CIFAR-10-LT, 7 percentage points (pp) increase relative to the CDB-CE Loss was recorded. There is more room for improvement on Places-LT.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2008.11a
• /
• pp.341-342
• /
• 2008

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.45 no.6
• /
• pp.309-315
• /
• 2019

Objectives: The aim of this study was to evaluate the validity of the existing classification and difficulty index of impacted mandibular third molars in clinical situations and propose a more practical classification system. Materials and Methods: This study included 204 impacted mandibular third molars in 154 patients; panoramic x-ray images were obtained before tooth extraction. Factors including age, sex, and pattern of impaction were investigated. All impacted third molars were classified and scored for spatial relationship (1-5 points), depth (1-4 points), and ramus relationship (1-3 points). All variables were measured twice by the same observer at a minimum interval of one month. Finally, the difficulty index was defined based on the total points scored as slightly difficult (3-4 points), moderately difficult (5-7 points), very difficult (8-10 points), and extremely difficult (11-12 points). Results: The strength of agreement of the total points scored and difficulty index were 0.855 and 0.746, respectively. Most cases were classified as moderately difficult (73.0%). Although only 13 out of 204 cases (6.4%) were classified as extremely difficult, patients classified as extremely difficult were the oldest (P<0.05). Conclusion: For difficulty classification, the authors propose one more difficult category beyond the existing three-step difficulty index: the clinician should consider the patient's age in the difficulty index evaluation.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.8
• /
• pp.315-324
• /
• 2022

In this paper, we propose KoEPT, a Transformer-based generative model for automatic math word problems solving. A math word problem written in human language which describes everyday situations in a mathematical form. Math word problem solving requires an artificial intelligence model to understand the implied logic within the problem. Therefore, it is being studied variously across the world to improve the language understanding ability of artificial intelligence. In the case of the Korean language, studies so far have mainly attempted to solve problems by classifying them into templates, but there is a limitation in that these techniques are difficult to apply to datasets with high classification difficulty. To solve this problem, this paper used the KoEPT model which uses 'expression' tokens and pointer networks. To measure the performance of this model, the classification difficulty scores of IL, CC, and ALG514, which are existing Korean mathematical sentence problem datasets, were measured, and then the performance of KoEPT was evaluated using 5-fold cross-validation. For the Korean datasets used for evaluation, KoEPT obtained the state-of-the-art(SOTA) performance with 99.1% in CC, which is comparable to the existing SOTA performance, and 89.3% and 80.5% in IL and ALG514, respectively. In addition, as a result of evaluation, KoEPT showed a relatively improved performance for datasets with high classification difficulty. Through an ablation study, we uncovered that the use of the 'expression' tokens and pointer networks contributed to KoEPT's state of being less affected by classification difficulty while obtaining good performance.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.11
• /
• pp.4016-4027
• /
• 2014

Understanding network latency is important for providing consistent and acceptable levels of services in network-based applications. However, due to the difficulty of estimating applications' network demands and the difficulty of network latency modeling the management of network resources has often been ignored. We expect that, since network latency repeats cycles of congested states, a systematic classification method for network status would be helpful to simplify issues in network resource managements. This paper presents a simple empirical method to classify network status with a real operational network. By observing oscillating behavior of end-to-end latency we determine networks' status in run time. Five typical network statuses are defined based on a long-term stability and a short-term burstiness. By investigating prediction accuracies of several simple numerical models we show the effectiveness of the network status classification. Experimental results show that around 80% reduction in prediction errors depending on network status.

• International Journal of Control, Automation, and Systems
• /
• v.2 no.1
• /
• pp.125-133
• /
• 2004

In this paper, we present a new method that combines the characteristics of edge in-formation and second-order neural networks for the classification of structural textures. The edges of a texture are extracted using an edge detection approach. From this edge information, classification features called second-order features are obtained. These features are fed into a second-order neural network for training and subsequent classification. It will be shown that the main disadvantage of using structural methods in texture classifications, namely, the difficulty of the extraction of texels, is overcome by the proposed method.

• Proceeding of Spring/Autumn Annual Conference of KHA
• /
• 2008.11a
• /
• pp.327-331
• /
• 2008

In Korea, welfare needs brought on by the rapidly growing of elderly population. Also, the senior group with spare time, wealth and health is increasing. Currently, the people have known about needs welfare system for improving quality of life during senescence personally and socially. In addition to the senior friendly industry, having distinction with simple approach medical services, will be great expanding after 2010. And there will be a new welfare services plan supporting long term care insurance for senior citizens. But we don't have prepared systematically organized senior friendly industry and products yet. So, the purpose of the study is to analyze systematically the classification characteristics of the senior friendly products according to the difficulty of daily living for senior citizens. Above all, the study finds out the senior friendly products characteristics according to the consumer's various situations about health and housing space. Therefore the result of the study reveals that the simulation guidelines for choosing the products depend on the customer's needs are essential and useful to improve senior friendly industry. And the post evaluation data of senior friendly products would be put to practical use in the industry.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.46 no.5
• /
• pp.328-334
• /
• 2020

Objectives: This study aimed to validate the effectiveness of a recently proposed difficulty index for removal of impacted mandibular third molars based on extraction time and suggest a modified difficulty index including the presence of pathologic conditions associated with third molars. Materials and Methods: This retrospective study enrolled 65 male patients younger than 25 years with third molars. Extraction time was calculated from start of the incision to the last suture. The difficulty scores for third molars were based on spatial relationship (1-5 points), depth (1-4 points), and ramus relationship (1-3 points) using cone-beam computed tomography. The difficulty index was defined as follows: I (3-4 points), II (5-7 points), III (8-10 points), and IV (11-12 points). The modified difficulty score was calculated by adding one point to the difficulty score if the third molar was associated with a pathologic condition. Two modified difficulty indices, based on the presence of pathologic conditions, were as follows: the half-level up difficulty index (HDI) and the one-level up difficulty index (ODI) from the recently proposed difficulty index. Results: The correlations between extraction time and difficulty index and or modified difficulty indices were significant (P<0.001). The correlation coefficient between extraction time and difficulty index was 0.584. The correlation coefficients between extraction time and HDI and ODI were 0.728 and 0.764, respectively. Conclusion: Extraction time of impacted third molars exhibited a moderate correlation with difficulty index and was strongly correlated with the modified indices. Considering the clinical implications, the difficulty index of surgical extraction should take into consideration the pathologic conditions associated with third molars.

• Proceedings of the KSR Conference
• /
• 2003.10b
• /
• pp.448-453
• /
• 2003

Most urban transit companies recognize the necessity of classification for facility management Classification for urban transit facility is necessary for standardization of maintenance management. The practical application. however. is not easy because of the absence of standardization of classification for urban transit facility and the difficulty in objectification of breakdown structure. This study suggests a proposal of classification for maintenance management in urban transit facility. This study defines standardization of classification as facility, work, maintenance and attribute to manage urban transit facility. And attribute classification consist of material, equipment and document. The suggested classification can be used as a useful maintenance management tool that enables evaluation of urban transit facility by standardization. The results of this study could be used as references for related urban transit companies.