• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.1324-1329
• /
• 2004

In this article we have tested stability of classifier based on fuzzy similarity in generalized Lukasiewicz structure. Two different tests for stability was made:In on test stability was checked respect to weight parameters and other test was carried out for idealvectors. Tests have made with three different classification problems.

• Journal of applied mathematics & informatics
• /
• v.19 no.1_2
• /
• pp.57-75
• /
• 2005

Quasigroups are algebraic structures closely related to Latin squares which have many different applications. There are several classifications of quasigroups based on their algebraic properties. In this paper we propose another classification based on the properties of strings obtained by specific quasigroup transformations. More precisely, in our research we identified some quasigroup transformations which can be applied to arbitrary strings to produce pseudo random sequences. We performed tests for randomness of the obtained pseudo-random sequences by random walks on torus. The randomness tests provided an empirical classification of quasigroups.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.6
• /
• pp.2468-2478
• /
• 2018

We propose the Sparse Feature Convolutional Neural Network (SFCNN) to reduce the volume of convolutional neural networks (CNNs). Despite the superior classification performance of CNNs, their enormous network volume requires high computational cost and long processing time, making real-time applications such as online-training difficult. We propose an advanced network that reduces the volume of conventional CNNs by producing a region-based sparse feature map. To produce the sparse feature map, two complementary region-based value extraction methods, cluster max extraction and local value extraction, are proposed. Cluster max is selected as the main function based on experimental results. To evaluate SFCNN, we conduct an experiment with two conventional CNNs. The network trains 59 times faster and tests 81 times faster than the VGG network, with a 1.2% loss of accuracy in multi-class classification using the Caltech101 dataset. In vehicle classification using the GTI Vehicle Image Database, the network trains 88 times faster and tests 94 times faster than the conventional CNNs, with a 0.1% loss of accuracy.

• Journal of Information Processing Systems
• /
• v.7 no.4
• /
• pp.717-732
• /
• 2011

Classification in biomedical applications is an important task that predicts or classifies an outcome based on a given set of input variables such as diagnostic tests or the symptoms of a patient. Traditionally the classification algorithms would have to digest a stationary set of historical data in order to train up a decision-tree model and the learned model could then be used for testing new samples. However, a new breed of classification called stream-based classification can handle continuous data streams, which are ever evolving, unbound, and unstructured, for instance--biosignal live feeds. These emerging algorithms can potentially be used for real-time classification over biosignal data streams like EEG and ECG, etc. This paper presents a pioneer effort that studies the feasibility of classification algorithms for analyzing biosignals in the forms of infinite data streams. First, a performance comparison is made between traditional and stream-based classification. The results show that accuracy declines intermittently for traditional classification due to the requirement of model re-learning as new data arrives. Second, we show by a simulation that biosignal data streams can be processed with a satisfactory level of performance in terms of accuracy, memory requirement, and speed, by using a collection of stream-mining algorithms called Optimized Very Fast Decision Trees. The algorithms can effectively serve as a corner-stone technology for real-time classification in future biomedical applications.

• Clinics in Shoulder and Elbow
• /
• v.21 no.2
• /
• pp.82-86
• /
• 2018

Background: Although a common shoulder disease, there are no accepted classification criteria for frozen shoulder (FS). This study therefore aimed to evaluate the accuracy of the conventionally used FS classification system. Methods: Primary FS patients (n=168) who visited our clinic from January 2010 to July 2015 were included in the study. After confirming restrictions of the glenohumeral joint motion and absence of history of systemic disease, trauma, shoulder surgery, shoulder muscle weakness, or specific x-ray abnormalities, the Zuckerman and Rokito's classification was employed for diagnosing primary FS. Following clinical diagnosis, each patient underwent a shoulder magnetic resonance imaging (MRI) and blood tests (lipid profile, glucose, hemoglobin A1c, and thyroid function). Based on the results of the blood tests and MRIs, the patients were reclassified, using the criteria proposed by Zuckerman and Rokito. Results: New diagnoses were ascertained including blood test results (16 patients with diabetes, 43 with thyroid abnormalities, and 149 with dyslipidemia), and MRI revealed intra-articular lesions in 81 patients (48.2%). After re-categorization based on the above findings, only 5 patients (3.0%) were classified having primary FS. The remaining 163 patients (97.0%) had either undiagnosed systemic or intrinsic abnormalities (89 patients), whereas 74 patients had both. Conclusions: These findings demonstrate that most patients clinically diagnosed with primary FS had undiagnosed systemic abnormalities and/or intra-articular pathologies. Therefore, a modification of the Zuckerman and Rokito's classification system for FS may be required to include the frequent combinations, rather than having a separate representation of systemic abnormalities and intrinsic causes.

• KSTLE International Journal
• /
• v.2 no.1
• /
• pp.35-39
• /
• 2001

In this paper, statistical approach is undertaken to investigate the classification of wear debris which is the key function of objective assessment of wear debris morphology. Wear tests are run to produce various kinds of wear debris. The images of wear debris from wear tests are captured with image acquisition equipment. By thresholding, two-dimensional binary images of wear debris are made and, then, morphological parameters are used to quantify the images of debris. Parametric and nonparametric discriminant method are employed to classify wear debris into predefined wear conditions. It is demonstrated that classification accuracy of parametric and nonparametric discriminant method is similar. The selected use of morphological parameters by stepwise discriminant analysis can generally improve the classification accuracy of parametric and nonparametric discriminant method.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2007.05a
• /
• pp.1142-1146
• /
• 2007

In this studies, subjective evaluation of heavy-weight floor impact sound through classification was conducted. Heavyweight impact sounds generated by an impact ball were recorded through dummy heads in apartment buildings. The recordings were classified according to the frequency characteristics of the floor impact sounds which are influenced by the floor structure with different boundary conditions and composite materials. The characteristics of the floor impact noise were investigated by paired comparison tests and semantic differential tests. Sound sources for auditory experiment were selected based on the actual noise levels with perceptual level differences. The results showed that roughness and fluctuation strength as well as loudness of the heavy-weight impact noise had a major effect on annoyance.

• Earthquakes and Structures
• /
• v.15 no.1
• /
• pp.1-8
• /
• 2018

A new site classification system and site coefficients based on local site conditions in Korea were developed and implemented as a part of minimum design load requirements for general seismic design. The new site classification system adopted bedrock depth and average shear wave velocity of soil above the bedrock as parameters for site classification. These code provisions were passed through a public hearing process before it was enacted. The public hearing process recommended to modify the naming of site classes and adjust the amplification factors so that the level of short-period amplification is suitable for economical seismic design. In this paper, the new code provisions were assessed using dynamic centrifuge tests and by comparing the design response spectra (DRS) with records from 2016 Gyeongju earthquake, the largest earthquake in history of instrumental seismic observation in Korea. The dynamic centrifuge tests were performed to simulate the representative Korean site conditions, such as shallow depth to bedrock and short-period amplification characteristics, and the results corroborated with the new DRS. The Gyeongju earthquake records also showed good agreement with the DRS. In summary, the new code provisions are reliable for representing the site amplification characteristic of shallow bedrock condition in Korea.

• Transactions on Control, Automation and Systems Engineering
• /
• v.2 no.1
• /
• pp.40-46
• /
• 2000

In this paper, a transient identification based on a Hidden Markov Model (HMM) has been suggested and evaluated experimentally for the classification of transients in the dynamic process. The transient can be identified by its unique time dependent patterns related to the principal variables. The HMM, a double stochastic process, can be applied to transient identification which is a spatial and temporal classification problem under a statistical pattern recognition framework. The HMM is created for each transient from a set of training data by the maximum-likelihood estimation method. The transient identification is determined by calculating which model has the highest probability for the given test data. Several experimental tests have been performed with normalization methods, clustering algorithms, and a number of states in HMM. Several experimental tests have been performed including superimposing random noise, adding systematic error, and untrained transients. The proposed real-time transient identification system has many advantages, however, there are still a lot of problems that should be solved to apply to a real dynamic process. Further efforts are being made to improve the system performance and robustness to demonstrate reliability and accuracy to the required level.

• Proceedings of the Earthquake Engineering Society of Korea Conference
• /
• 2005.03a
• /
• pp.90-101
• /
• 2005

For the site characterization at two representative inland areas, Gyeongju and Hongsung, in Korea, in-situ seismic tests containing boring investigations and resonant column tests were performed and site-specific ground response analyses were conducted using equivalent linear as well as nonlinear scheme. The soil deposits in Korea were shallower and stiffer than those in the western US, from which the site classification system and site coefficients in Korea were derived. Most sites were categorized as site classes C and D based on the mean shear wave velocity to 30 m, Vs30 ranging between 250 and 650 m/s. Based on the acceleration response spectra determined from the site-specific analyses, the site coefficients specified in the Korean seismic design guide underestimate the ground motion in the short-period band and overestimate the ground motion in mid-period band. These differences can be explained by the differences in the bedrock depth and the soil stiffness profile between Korea and western US. The site coefficients were re-evaluated and the preliminary site classification system was introduced accounting for the local geologic conditions on the Korean peninsula.