• ETRI Journal
• /
• v.42 no.5
• /
• pp.761-772
• /
• 2020

This paper presents an automatic proficiency assessment method for a non-native Korean read utterance using bidirectional long short-term memory (BLSTM)-based acoustic models (AMs) and speech data augmentation techniques. Specifically, the proposed method considers two scenarios, with and without prompted text. The proposed method with the prompted text performs (a) a speech feature extraction step, (b) a forced-alignment step using a native AM and non-native AM, and (c) a linear regression-based proficiency scoring step for the five proficiency scores. Meanwhile, the proposed method without the prompted text additionally performs Korean speech recognition and a subword un-segmentation for the missing text. The experimental results indicate that the proposed method with prompted text improves the performance for all scores when compared to a method employing conventional AMs. In addition, the proposed method without the prompted text has a fluency score performance comparable to that of the method with prompted text.

• Korean Journal of Remote Sensing
• /
• v.24 no.1
• /
• pp.79-89
• /
• 2008

Irregular temporal sampling is a common feature of geophysical and biological time series in remote sensing. This study proposes an on-line system for reconstructing observation image series contaminated by noises resulted from mechanical problems or sensing environmental condition. There is also a high likelihood that during the data acquisition periods the target site corresponding to any given pixel may be covered by fog or cloud, thereby resulting in bad or missing observation. The surface parameters associated with the land are usually dependent on the climate, and many physical processes that are displayed in the image sensed from the land then exhibit temporal variation with seasonal periodicity. A feedback system proposed in this study reconstructs a sequence of images remotely sensed from the land surface having the physical processes with seasonal periodicity. The harmonic model is used to track seasonal variation through time, and a Gibbs random field (GRF) is used to represent the spatial dependency of digital image processes. The experimental results of this simulation study show the potentiality of the proposed system to reconstruct the image series observed by imperfect sensing technology from the environment which are frequently influenced by bad weather. This study provides fundamental information on the elements of the proposed system for right usage in application.

• Journal of Information Processing Systems
• /
• v.17 no.4
• /
• pp.754-771
• /
• 2021

In the task of continuous dimension emotion recognition, the parts that highlight the emotional expression are not the same in each mode, and the influences of different modes on the emotional state is also different. Therefore, this paper studies the fusion of the two most important modes in emotional recognition (voice and visual expression), and proposes a two-mode dual-modal emotion recognition method combined with the attention mechanism of the improved AlexNet network. After a simple preprocessing of the audio signal and the video signal, respectively, the first step is to use the prior knowledge to realize the extraction of audio characteristics. Then, facial expression features are extracted by the improved AlexNet network. Finally, the multimodal attention mechanism is used to fuse facial expression features and audio features, and the improved loss function is used to optimize the modal missing problem, so as to improve the robustness of the model and the performance of emotion recognition. The experimental results show that the concordance coefficient of the proposed model in the two dimensions of arousal and valence (concordance correlation coefficient) were 0.729 and 0.718, respectively, which are superior to several comparative algorithms.

• The Journal of Bigdata
• /
• v.8 no.1
• /
• pp.35-47
• /
• 2023

It is crucial to develop effective and efficient big data analytics methods for problem-solving in the field of business in order to improve the performance of data analytics and reduce costs and risks in the analysis of customer data. In this study, a big data-driven data analysis system using artificial intelligence techniques is designed to increase the accuracy of big data analytics along with the rapid growth of the field of data science. We present a key direction for big data analysis systems through missing value imputation, outlier detection, feature extraction, utilization of explainable artificial intelligence techniques, and exploratory data analysis. Our objective is not only to develop big data analysis techniques with complex structures of business data but also to bridge the gap between the theoretical ideas in artificial intelligence methods and the analysis of real-world data in the field of business.

• Geophysics and Geophysical Exploration
• /
• v.20 no.2
• /
• pp.78-87
• /
• 2017

For efficient data processing, trace interpolation and regularization techniques should be antecedently applied to the seismic data which were irregularly sampled with missing traces. Among many interpolation techniques, MWNI (Minimum Weighted Norm Interpolation) technique is one of the most versatile techniques and widely used to regularize seismic data because of easy extension to the high-order module and low computational cost. However, since it is difficult to interpolate spatially aliased data using this technique, model-constrained MWNI was suggested to compensate for this problem. In this paper, conventional MWNI and model-constrained MWNI modules have been developed in order to analyze their performance using synthetic data and validate the applicability to the field data. The result by using model-constrained MWNI was better in spatially aliased data. In order to verify the applicability to the field data, interpolation and regularization were performed for two field data sets, respectively. Firstly, the seismic data acquired in Ulleung Basin gas hydrate field was interpolated. Even though the data has very chaotic feature and complex structure due to the chimney, the developed module showed fairly good interpolation result. Secondly, very irregularly sampled and widely missing seismic data was regularized and the connectivity of events was quite improved. According to these experiments, we can confirm that the developed module can successfully interpolate and regularize the irregularly sampled field data.

• The Journal of Korean Academy of Prosthodontics
• /
• v.17 no.1
• /
• pp.35-46
• /
• 1979

A number of experimental studies have been carried out in order to clarify the question as to how temporomandibular joint adapt to the changes of mandibular movement and occlusal equilibration. Recently, the studies on the interrelations between anatomical structure of temporomandibular joint and the state of occlusion have been actively performed in dentistry particularly in prosthodontic field. Author performed extraction of unilateral mandibular molars in 30 mature male rats, and observed histological changes of temporomandibular joint through the light microscope. Following results were obtained. 1. The loss of unilateral teeth gave rise to the changes in the location of condylar head, that is, interior displacement of condylar head in the extraction side and upper displacement in the non-extraction side. 2. Articular disk was compressed by the interior surface of condylar head, resulting in its extension below the condylar neck in the extraction side, and the histological arrangement of the compressed area showed irregular feature. 3. The extension of articular disk below the condylar neck was accompanied with the contraction of muscle fibers which were originated from the articular disk. 4. The cartilage layer of articular fossa to the exterior of the extraction side showed hypertropy. 5. Early in the experiment, the inernal extremity of condylar head of extract ion side showed bone resorption, and cartilage layer of condylar head showed hypertropy. At 12 weeks after experiment, the condylar surface showed flattened, and the cartilage layer of condylar head was replaced by the compact bone. 6. The articular disk showed the formation of pannus in the extraction side as well as in the non-extraction side. 7. The occlusal disturbance due to unilateral missing teeth has brought about the non-inflammatory retrogressive change and osteoarthrotic change late in the experiment.

• The KIPS Transactions:PartB
• /
• v.12B no.6 s.102
• /
• pp.697-702
• /
• 2005

The focused crawler is a topic-driven document-collecting crawler that was suggested as a promising alternative of maintaining up-to-date web document Indices in search engines. A major problem inherent in previous focused crawlers is the liability of missing highly relevant documents that are linked from off-topic documents. This problem mainly originated from the lack of consideration of structural information in a document. Traditional weighting method such as TFIDF employed in document classification can lead to this problem. In order to improve the performance of focused crawlers, this paper proposes a scheme of locality-based document segmentation to determine the relevance of a document to a specific topic. We segment a document into a set of sub-documents using contextual features around the hyperlinks. This information is used to determine whether the crawler would fetch the documents that are linked from hyperlinks in an off-topic document.

• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.18
• /
• pp.8145-8147
• /
• 2016

Background: Atypical squamous cells of undetermined significance (ASCUS) feature a wide variety of cervical cells, including benign and malignant examples. The management of ASCUS is complicated. Guidelines for office gynecology in Japan recommend performing a high-risk human papillomavirus (HPV) test as a rule. The guidelines also recommend repeat cervical cytology after 6 and 12 months, or immediate colposcopy. The purpose of this study was to determine the clinical significance of ASCUS. Materials and Methods: Between January 2012 and December 2014, a total of 162 patients underwent cervical conization for cervical intraepithelial neoplasia grade 3 (CIN3), carcinoma in situ, squamous cell carcinoma, microinvasive squamous cell carcinoma, and adenocarcinoma in situ at our hospital. The results of cervical cytology prior to conization, the pathology after conization, and high-risk HPV testing were obtained from clinical records and analyzed retrospectively. Results: Based on cervical cytology, 31 (19.1%) of 162 patients were primarily diagnosed with ASCUS. Among these, 25 (80.6%) were positive for high-risk HPV, and the test results of the remaining 6 patients (19.4%) were uncertain. In the final pathological diagnosis after conization, 27 (87.1%) and 4 patients (12.9%) were diagnosed with CIN3 and carcinoma in situ, respectively. Conclusions: Although ASCUS is known as a low-risk abnormal cervical cytology, approximately 20% of patients who underwent cervical conization had ASCUS. The relationship between the cervical cytology of ASCUS and the final pathological results for CIN3 or invasive carcinoma should be investigated statistically. In cases of ASCUS, we recommend HPV tests or colposcopic examination rather than cytological follow-up, because of the risk of missing CIN3 or more advanced disease.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.41 no.2
• /
• pp.103-109
• /
• 2017

Metal 3D printing, which is an additive metal manufacturing (AMM) process, enables the development of full-density metallic tools and parts using metal powders that are precisely delivered and controlled for deposition with no powder bed. However, some unknown geometric defects and irregular geometric features on an STL model can possibly result in incorrect metal part fabrication after the build. This study first proposes a methodical approach for verifying the build part, including the missing facet problems in an STL model, by defining some irregular features that possibly exist on the part. Second, 2D tool paths on each build layer were investigated for detecting any singular region inside the layer. The method was implemented for building two sample STL models using a direct energy deposition process, and finally, it was visually simulated for diagnosis.

• Journal of Oral Medicine and Pain
• /
• v.24 no.1
• /
• pp.81-94
• /
• 1999

This study intends to find usefulness of the computer assisted dental identification in mass disaster. The variety of dental characteristics was investigated through the research of dental records of 508 adults. And a computer assisted simulation program was used to evaluate the selectivity of dental identification. Findings were as follows : 1. Combinations of dental characteristics were found 155 types. The most various dental characteristic was showed on the mandibular first molar. 99.0% of subjects had dental characteristics for dental identification. 2. The posterior teeth, in comparison with anterior teeth, showed higher selectivity in dental identification which was enhanced by information on the material, type and cavity of the restoration. 3. The variety of dental characteristics was mainly found on the combinations of missing tooth with the material, type and cavity of the restoration. 4. The computer assisted dental identification program, with informa- tion about one's tooth state, made individual identification possible when there was only a part of the teeth in a corpse. 5. The computer assisted dental identification had considerably high selectivity based on the variety of dental characteristics. And it was also performed faster and preciser than the existing identification methods. Based on the results of this study, there are various combinations of the feature of the tooth itself with dental characteristics caused by a certain type of treatment on the teeth. And using the computer assisted dental identification program based on this, dental identification can be more efficient economically and more useful than any other forensic identification methods.