• Journal of Intelligence and Information Systems
• /
• v.27 no.3
• /
• pp.57-73
• /
• 2021

Maintenance and prevention of failure through anomaly detection of ICT infrastructure is becoming important. System monitoring data is multidimensional time series data. When we deal with multidimensional time series data, we have difficulty in considering both characteristics of multidimensional data and characteristics of time series data. When dealing with multidimensional data, correlation between variables should be considered. Existing methods such as probability and linear base, distance base, etc. are degraded due to limitations called the curse of dimensions. In addition, time series data is preprocessed by applying sliding window technique and time series decomposition for self-correlation analysis. These techniques are the cause of increasing the dimension of data, so it is necessary to supplement them. The anomaly detection field is an old research field, and statistical methods and regression analysis were used in the early days. Currently, there are active studies to apply machine learning and artificial neural network technology to this field. Statistically based methods are difficult to apply when data is non-homogeneous, and do not detect local outliers well. The regression analysis method compares the predictive value and the actual value after learning the regression formula based on the parametric statistics and it detects abnormality. Anomaly detection using regression analysis has the disadvantage that the performance is lowered when the model is not solid and the noise or outliers of the data are included. There is a restriction that learning data with noise or outliers should be used. The autoencoder using artificial neural networks is learned to output as similar as possible to input data. It has many advantages compared to existing probability and linear model, cluster analysis, and map learning. It can be applied to data that does not satisfy probability distribution or linear assumption. In addition, it is possible to learn non-mapping without label data for teaching. However, there is a limitation of local outlier identification of multidimensional data in anomaly detection, and there is a problem that the dimension of data is greatly increased due to the characteristics of time series data. In this study, we propose a CMAE (Conditional Multimodal Autoencoder) that enhances the performance of anomaly detection by considering local outliers and time series characteristics. First, we applied Multimodal Autoencoder (MAE) to improve the limitations of local outlier identification of multidimensional data. Multimodals are commonly used to learn different types of inputs, such as voice and image. The different modal shares the bottleneck effect of Autoencoder and it learns correlation. In addition, CAE (Conditional Autoencoder) was used to learn the characteristics of time series data effectively without increasing the dimension of data. In general, conditional input mainly uses category variables, but in this study, time was used as a condition to learn periodicity. The CMAE model proposed in this paper was verified by comparing with the Unimodal Autoencoder (UAE) and Multi-modal Autoencoder (MAE). The restoration performance of Autoencoder for 41 variables was confirmed in the proposed model and the comparison model. The restoration performance is different by variables, and the restoration is normally well operated because the loss value is small for Memory, Disk, and Network modals in all three Autoencoder models. The process modal did not show a significant difference in all three models, and the CPU modal showed excellent performance in CMAE. ROC curve was prepared for the evaluation of anomaly detection performance in the proposed model and the comparison model, and AUC, accuracy, precision, recall, and F1-score were compared. In all indicators, the performance was shown in the order of CMAE, MAE, and AE. Especially, the reproduction rate was 0.9828 for CMAE, which can be confirmed to detect almost most of the abnormalities. The accuracy of the model was also improved and 87.12%, and the F1-score was 0.8883, which is considered to be suitable for anomaly detection. In practical aspect, the proposed model has an additional advantage in addition to performance improvement. The use of techniques such as time series decomposition and sliding windows has the disadvantage of managing unnecessary procedures; and their dimensional increase can cause a decrease in the computational speed in inference.The proposed model has characteristics that are easy to apply to practical tasks such as inference speed and model management.

• Proceedings of the Materials Research Society of Korea Conference
• /
• 2011.05a
• /
• pp.5-5
• /
• 2011

The research and development of hybrid electric vehicle (HEV), plug-in hybrid electric vehicle (PHEV) and electric vehicle (EV) are intensified due to the energy crisis and environmental concerns. In order to meet the challenging requirements of powering HEV, PHEV and EV, the current lithium battery technology needs to be significantly improved in terms of the cost, safety, power and energy density, as well as the calendar and cycle life. One new technology being developed is the utilization of composite cathode by mixing two different types of insertion compounds [e.g., spinel $LiMn_2O_4$ and layered $LiMO_2$ (M=Ni, Co, and Mn)]. Recently, some studies on mixing two different types of cathode materials to make a composite cathode have been reported, which were aimed at reducing cost and improving self-discharge. Numata et al. reported that when stored in a sealed can together with electrolyte at $80^{\circ}C$ for 10 days, the concentrations of both HF and $Mn^{2+}$ were lower in the can containing $LiMn_2O_4$ blended with $LiNi_{0.8}Co_{0.2}O_2$ than that containing $LiMn_2O_4$ only. That reports clearly showed that this blending technique can prevent the decline in capacity caused by cycling or storage at elevated temperatures. However, not much work has been reported on the charge-discharge characteristics and related structural phase transitions for these composite cathodes. In this presentation, we will report our in situ x-ray diffraction studies on this mixed composite cathode material during charge-discharge cycling. The mixed cathodes were incorporated into in situ XRD cells with a Li foil anode, a Celgard separator, and a 1M $LiPF_6$ electrolyte in a 1 : 1 EC : DMC solvent (LP 30 from EM Industries, Inc.). For in situ XRD cell, Mylar windows were used as has been described in detail elsewhere. All of these in situ XRD spectra were collected on beam line X18A at National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory using two different detectors. One is a conventional scintillation detector with data collection at 0.02 degree in two theta angle for each step. The other is a wide angle position sensitive detector (PSD). The wavelengths used were 1.1950 ${\AA}$ for the scintillation detector and 0.9999 A for the PSD. The newly installed PSD at beam line X18A of NSLS can collect XRD patterns as short as a few minutes covering $90^{\circ}$ of two theta angles simultaneously with good signal to noise ratio. It significantly reduced the data collection time for each scan, giving us a great advantage in studying the phase transition in real time. The two theta angles of all the XRD spectra presented in this paper have been recalculated and converted to corresponding angles for ${\lambda}=1.54\;{\AA}$, which is the wavelength of conventional x-ray tube source with Cu-$k{\alpha}$ radiation, for easy comparison with data in other literatures. The structural changes of the composite cathode made by mixing spinel $LiMn_2O_4$ and layered $Li-Ni_{1/3}Co_{1/3}Mn_{1/3}O_2$ in 1 : 1 wt% in both Li-half and Li-ion cells during charge/discharge are studied by in situ XRD. During the first charge up to ~5.2 V vs. $Li/Li^+$, the in situ XRD spectra for the composite cathode in the Li-half cell track the structural changes of each component. At the early stage of charge, the lithium extraction takes place in the $LiNi_{1/3}Co_{1/3}Mn_{1/3}O_2$ component only. When the cell voltage reaches at ~4.0 V vs. $Li/Li^+$, lithium extraction from the spinel $LiMn_2O_4$ component starts and becomes the major contributor for the cell capacity due to the higher rate capability of $LiMn_2O_4$. When the voltage passed 4.3 V, the major structural changes are from the $LiNi_{1/3}Co_{1/3}Mn_{1/3}O_2$ component, while the $LiMn_2O_4$ component is almost unchanged. In the Li-ion cell using a MCMB anode and a composite cathode cycled between 2.5 V and 4.2 V, the structural changes are dominated by the spinel $LiMn_2O_4$ component, with much less changes in the layered $LiNi_{1/3}Co_{1/3}Mn_{1/3}O_2$ component, comparing with the Li-half cell results. These results give us valuable information about the structural changes relating to the contributions of each individual component to the cell capacity at certain charge/discharge state, which are helpful in designing and optimizing the composite cathode using spinel- and layered-type materials for Li-ion battery research. More detailed discussion will be presented at the meeting.

• Journal of the korean academy of Pediatric Dentistry
• /
• v.34 no.4
• /
• pp.569-578
• /
• 2007

This study was performed to compare the shear bond strength of self etching system and two bottle bonding system with or without laser preparation. Group I was prepared with high speed rotary instrument and $Prompt^{TM}$ L-$Pop^{TM}$, group II with Er:YAG laser and $Prompt^{TM}$ L-$Pop^{TM}$, group III with Er:YAG laser, 37% phosphoric acid and Single bond, group IV with Er:YAG laser and Single bond and group V with high speed, etching and Single bond. And also observation of the prepared and etched dentin surface were performed under scanning electro-microscope. The possibility of clinical application of laser preparation which might have an advantage to reduce pain for children with less unfavorable noise were evaluated. The results obtained are as follows; 1. Group V showed significantly higher bond strength than other groups. And group IV showed significantly lower bond strength than other groups. 2. There was no significant difference between group I and group III. 3. Group II showed significantly lower bond strength than group I, III, V, but showed significantly higher bond strength than group IV. 4. Under scanning electro-microscope, laser-preparated dentin surface showed high irregularity and no smear layer. The surface showed less irregularities and more exposed dentinal tubules with etching. Laser preparation has many advantages over conventional tooth preparation. But this method showed lower resin bonding strength. Laser preparated tooth surface differed from the conventionally preparated tooth surface. More researches are needed on suitable methods for laser preparated dentin surface.

• Journal of the Korean Society of Radiology
• /
• v.13 no.4
• /
• pp.581-588
• /
• 2019

MRI is superior when contrasted to help the organization generate artifacts resolution, but also affect the diagnosis and create a image that can not be read. Metal is inserted into the tooth, it is necessary to often be inhibited in imaging by causing the geometric distortion due to the majority and if the difference between the magnetic susceptibility of a ferromagnetic material or paramagnetic reducing them. The purpose of this study is to conduct a metal artefact in accordance with the analysis using a diamagnetic material. The magnetic material include a wire for the orthodontic bracket and a stainless steel was used as a diamagnetic material was used copper, zinc, bismuth. Testing equipment is sequenced using 1.5T, 3T was used was measured using a SE, TSE, GE, EPI. A self-produced phantom material was used for agarose gel (10%) to a uniform signal artifacts causing materials are stainless steel were tested by placing in the center of the phantom and cover inspection of the positive cube diamagnetic material of 10mm each length.After a measurement artefact artifact zone settings area was calculated using the Wand tool After setting the Low Threshold value of 10 in the image obtained by subtracting images, including magnetic material from a pure tool phantom images using Image J. Metal artifacts occur in stainless steel metal artifact reduction was greatest in the image with the bismuth diamagnetic materials of copper and zinc is slightly reduced, but the difference in degree will not greater. The reason for this is thought to be due to hayeotgi offset most of the susceptibility in bismuth diamagnetic susceptibility of most small ferromagnetic. Most came with less artifacts in image of bismuth in both 1.5T and 3T. Sequence-specific artifact reduction was most reduced artifacts from the TSE 1.5T 3T was reduced in the most artifacts from SE. Signal-to-noise ratio was the lowest SNR is low, appears in the implant, the 1.5T was the Implant + Bi Cu and Zn showed similar results to each other. Therefore, the results of artifacts variation of diamagnetic material, magnetic susceptibility (${\chi}$) is the most this shows the reduced aspect lower than the implant artificial metal artifacts criteria in the video using low bismuth susceptibility to low material the more metal artifacts It was found that the decrease. Therefore, based on the study on the increase, the metal artifacts reduction for the whole, as well as dental prosthesis future orthodontic materials in a way that can even reduce the artifact does not appear which has been pointed out as a disadvantage of the solutions of conventional metal artifact It is considered to be material.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.173-198
• /
• 2020

For a long time, many studies have been conducted on predicting the success of campaigns for customers in academia, and prediction models applying various techniques are still being studied. Recently, as campaign channels have been expanded in various ways due to the rapid revitalization of online, various types of campaigns are being carried out by companies at a level that cannot be compared to the past. However, customers tend to perceive it as spam as the fatigue of campaigns due to duplicate exposure increases. Also, from a corporate standpoint, there is a problem that the effectiveness of the campaign itself is decreasing, such as increasing the cost of investing in the campaign, which leads to the low actual campaign success rate. Accordingly, various studies are ongoing to improve the effectiveness of the campaign in practice. This campaign system has the ultimate purpose to increase the success rate of various campaigns by collecting and analyzing various data related to customers and using them for campaigns. In particular, recent attempts to make various predictions related to the response of campaigns using machine learning have been made. It is very important to select appropriate features due to the various features of campaign data. If all of the input data are used in the process of classifying a large amount of data, it takes a lot of learning time as the classification class expands, so the minimum input data set must be extracted and used from the entire data. In addition, when a trained model is generated by using too many features, prediction accuracy may be degraded due to overfitting or correlation between features. Therefore, in order to improve accuracy, a feature selection technique that removes features close to noise should be applied, and feature selection is a necessary process in order to analyze a high-dimensional data set. Among the greedy algorithms, SFS (Sequential Forward Selection), SBS (Sequential Backward Selection), SFFS (Sequential Floating Forward Selection), etc. are widely used as traditional feature selection techniques. It is also true that if there are many risks and many features, there is a limitation in that the performance for classification prediction is poor and it takes a lot of learning time. Therefore, in this study, we propose an improved feature selection algorithm to enhance the effectiveness of the existing campaign. The purpose of this study is to improve the existing SFFS sequential method in the process of searching for feature subsets that are the basis for improving machine learning model performance using statistical characteristics of the data to be processed in the campaign system. Through this, features that have a lot of influence on performance are first derived, features that have a negative effect are removed, and then the sequential method is applied to increase the efficiency for search performance and to apply an improved algorithm to enable generalized prediction. Through this, it was confirmed that the proposed model showed better search and prediction performance than the traditional greed algorithm. Compared with the original data set, greed algorithm, genetic algorithm (GA), and recursive feature elimination (RFE), the campaign success prediction was higher. In addition, when performing campaign success prediction, the improved feature selection algorithm was found to be helpful in analyzing and interpreting the prediction results by providing the importance of the derived features. This is important features such as age, customer rating, and sales, which were previously known statistically. Unlike the previous campaign planners, features such as the combined product name, average 3-month data consumption rate, and the last 3-month wireless data usage were unexpectedly selected as important features for the campaign response, which they rarely used to select campaign targets. It was confirmed that base attributes can also be very important features depending on the type of campaign. Through this, it is possible to analyze and understand the important characteristics of each campaign type.