• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.43-62
• /
• 2019

At one time, the anomaly detection sector dominated the method of determining whether there was an abnormality based on the statistics derived from specific data. This methodology was possible because the dimension of the data was simple in the past, so the classical statistical method could work effectively. However, as the characteristics of data have changed complexly in the era of big data, it has become more difficult to accurately analyze and predict the data that occurs throughout the industry in the conventional way. Therefore, SVM and Decision Tree based supervised learning algorithms were used. However, there is peculiarity that supervised learning based model can only accurately predict the test data, when the number of classes is equal to the number of normal classes and most of the data generated in the industry has unbalanced data class. Therefore, the predicted results are not always valid when supervised learning model is applied. In order to overcome these drawbacks, many studies now use the unsupervised learning-based model that is not influenced by class distribution, such as autoencoder or generative adversarial networks. In this paper, we propose a method to detect anomalies using generative adversarial networks. AnoGAN, introduced in the study of Thomas et al (2017), is a classification model that performs abnormal detection of medical images. It was composed of a Convolution Neural Net and was used in the field of detection. On the other hand, sequencing data abnormality detection using generative adversarial network is a lack of research papers compared to image data. Of course, in Li et al (2018), a study by Li et al (LSTM), a type of recurrent neural network, has proposed a model to classify the abnormities of numerical sequence data, but it has not been used for categorical sequence data, as well as feature matching method applied by salans et al.(2016). So it suggests that there are a number of studies to be tried on in the ideal classification of sequence data through a generative adversarial Network. In order to learn the sequence data, the structure of the generative adversarial networks is composed of LSTM, and the 2 stacked-LSTM of the generator is composed of 32-dim hidden unit layers and 64-dim hidden unit layers. The LSTM of the discriminator consists of 64-dim hidden unit layer were used. In the process of deriving abnormal scores from existing paper of Anomaly Detection for Sequence data, entropy values of probability of actual data are used in the process of deriving abnormal scores. but in this paper, as mentioned earlier, abnormal scores have been derived by using feature matching techniques. In addition, the process of optimizing latent variables was designed with LSTM to improve model performance. The modified form of generative adversarial model was more accurate in all experiments than the autoencoder in terms of precision and was approximately 7% higher in accuracy. In terms of Robustness, Generative adversarial networks also performed better than autoencoder. Because generative adversarial networks can learn data distribution from real categorical sequence data, Unaffected by a single normal data. But autoencoder is not. Result of Robustness test showed that he accuracy of the autocoder was 92%, the accuracy of the hostile neural network was 96%, and in terms of sensitivity, the autocoder was 40% and the hostile neural network was 51%. In this paper, experiments have also been conducted to show how much performance changes due to differences in the optimization structure of potential variables. As a result, the level of 1% was improved in terms of sensitivity. These results suggest that it presented a new perspective on optimizing latent variable that were relatively insignificant.

• Microbiology and Biotechnology Letters
• /
• v.43 no.3
• /
• pp.267-274
• /
• 2015

Biopharmaceuticals and the cell substrates used for their manufacture are currently tested for porcine adventitious viruses due to the widespread use of porcine trypsin in cell culture. Porcine transmissible gastroenteritis virus (PTGV) is one of the major adventitious porcine viruses causing contaminated during the manufacture of biopharmaceuticals. Therefore, rapid and sensitive detection of PTGV is essential in ensuring the safety of biopharmaceuticals. A TaqMan probe real-time RT-PCR method was developed for the quantitative detection of PTGV contamination in cell substrates, raw materials, manufacturing processes, and final products, as well as PTGV clearance validation. Specific primers for the amplification of PTGV RNA were selected, and PTGV RNA was quantified by use of a specific TaqMan probe. Specificity, limit of detection (LOD), and robustness of the method was validated according to international guidelines on the validation of nucleic acid amplification tests. The sensitivity of the assay was calculated to be 1.10 × 10⁰ TCID₅₀/ml. The real-time RT-PCR method was validated to be reproducible, very specific to PTGV, and robust. The established real-time RT-PCR assay was successfully applied to the validation of Chinese Hamster Ovary (CHO)-K1 cells artificially infected with PTGV.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1032-1032
• /
• 2001

Artificial Neural Network (ANN) calibration techniques have been used commercially for agricultural applications since the mid-nineties. Global models, based on transmission data from 850 to 1050 nm, are used routinely to measure protein and moisture in wheat and barley and also moisture in triticale, rye, and oats. These models are currently used commercially in approx. 15 countries throughout the world. Results concerning earlier European ANN models are being published elsewhere. Some of the findings from that study will be discussed here. ANN models have also been developed for coarsely ground samples of compound feed and feed ingredients, again measured in transmission mode from 850 to 1050 nm. The performance of models for pig- and poultry feed will be discussed briefly. These models were developed from a very large data set (more than 20,000 records), and cover a very broad range of finished products. The prediction curves are linear over the entire range for protein, fat moisture, fibre, and starch (measured only on poultry feed), and accuracy is in line with the performance of smaller models based on Partial Least Squares (PLS). A simple bias adjustment is sufficient for calibration transfer across instruments. Recently, we have investigated the possible use of ANN for a different type of NIR spectrometer, based on reflectance data from 1100 to 2500 nm. In one study, based on data for protein, fat, and moisture measured on unground compound feed samples, dedicated ANN models for specific product classes (cattle feed, pig feed, broiler feed, and layers feed) gave moderately better Standard Errors of Prediction (SEP) compared to modified PLS (MPLS). However, if the four product classes were combined into one general calibration model, the performance of the ANN model deteriorated only slightly compared to the class-specific models, while the SEP values for the MPLS predictions doubled. Brix value in molasses is a measure of sugar content. Even with a huge dataset, PLS models were not sufficiently accurate for commercial use. In contrast an ANN model based on the same data improved the accuracy considerably and straightened out non-linearity in the prediction plot. The work of Mr. David Funk (GIPSA, U. S. Department of Agriculture) who has studied the influence of various types of spectral distortions on ANN- and PLS models, thereby providing comparative information on the robustness of these models towards instrument differences, will be discussed. This study was based on data from different classes of North American wheat measured in transmission from 850 to 1050 nm. The distortions studied included the effect of absorbance offset pathlength variation, presence of stray light bandwidth, and wavelength stretch and offset (either individually or combined). It was shown that a global ANN model was much less sensitive to most perturbations than class-specific GIPSA PLS calibrations. It is concluded that ANN models based on large data sets offer substantial advantages over PLS models with respect to accuracy, range of materials that can be handled by a single calibration, stability, transferability, and sensitivity to perturbations.

• Korean Journal of Remote Sensing
• /
• v.30 no.1
• /
• pp.47-59
• /
• 2014

Volcanic ash (VA) can be estimated by remote sensing sensors through their spectral signatures determined by the inherent optical property (IOP) including complex refractive index and the scattering properties. Until now, a very limited range of VA refractive indices has been reported and the VA from each volcanic eruption has a different composition. To improve the robustness of VA remote sensing, there is a need to understanding of VA - radiation interactions. In this study, we calculated extinction coefficient, scattering phase function, asymmetry factor, and single scattering albedo which show different values between andesite and pumice. Then, IOPs were used to analyze the relationship between theoretical remote sensed radiation calculated by radiative transfer model under various aerosol optical thickness (${\tau}$) and sun-sensor geometries and characteristics of VA. It was found that the mean rate of change of radiance at top of atmosphere versus ${\tau}$ is six times larger than in radiance values at 0.55 ${\mu}m$. At the surface, positive correlation dominates when ${\tau}$ <1, but negative correlation dominates when ${\tau}$ >1. However, radiance differences between andesite and pumice at 11 ${\mu}m$ are very small. These differences between two VA types are expressed as the polynomial regression functions and that increase as VA optical thickness increases. Finally, these results would allow VA to be better characterized by remote sensing sensors.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.36 no.7
• /
• pp.805-811
• /
• 2012

The reliability based design optimization (RBDO) approach requires high computing cost to consider uncertainties. In order to reduce the design cost, the single loop single vector (SLSV) approach has been developed for RBDO. This method can reduce the cost in calculating deign sensitivity by elimination of the nested optimization process. However, this process causes the increment of the instability or inaccuracy of the method according to the problem characteristics. Therefore, the method may not give accurate solution or the robustness of the solution is not guaranteed. Especially, when the function is concave, the process frequently diverges. In this research, the concept of the conjugate gradient method for unconstrained optimization is utilized to develop a new single loop single vector method. The conjugate gradient is calculated with gradient directions at the most probable points (MPP) of previous cycles. Mathematical examples are solved for the verification of the proposed method. The numeri cal performances of the obtained results are compared to those of other RBDO methods. The SLSV approach using conjugate gradient is not greatly influenced by the problem characteristics and improves its convergence capability.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.2
• /
• pp.64-70
• /
• 2011

Portable artificial electronic nose (E-nose) system suffers from noisy fluctuation in surroundings such as temperature, vapor concentration, and gas flow, because its measuring condition is not controled precisely as in the laboratory. It is important to develop a simple and robust vapor recognition technique applicable to this uncontrolled measurement, especially for the portable measuring and diagnostic system which are expanding its area with the improvements in micro bio sensor technology. This study used a PDA-based portable E-nose to collect the uncontrolled vapor measurement signals, and applied the image matching algorithm developed in the previous study on the measured signal to verify its robustness and improved accuracy in portable vapor recognition. The results showed not only its consistent performance under noisy fluctuation in the portable measurement signal, but also an advanced recognition accuracy for 2 similar vapor species which have been hard to discriminate with the conventional maximum sensitivity feature extraction method. The proposed method can be easily applied to the data processing of the ubiquitous sensor network (USN) which are usually exposed to various operating conditions. Furthermore, it will greatly help to realize portable medical diagnostic and environment monitoring system with its robust performance and high accuracy.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.20 no.6
• /
• pp.751-759
• /
• 2007

We have developed several methods for the optimization problem having large-scale and highly nonlinear system. First, step by step method in optimization process was employed to improve the convergence. In addition, techniques of furnishing good initial guesses for analysis using sensitivity information acquired from optimization iteration, and of manipulating analysis/optimization convergency criterion motivated from simultaneous technique were used. We applied them to flow control problem and verified their efficiency and robustness. However, they are based on quasi-Newton method that approximate the Hessian matrix using exact first derivatives. However solution of the Navier-Stokes equations are very cost, so we want to improve the efficiency of the optimization algorithm as much as possible. Thus we develop a true Newton method that uses exact Hessian matrix. And we apply that to the three-dimensional problem of flow around a sphere. This problem is certainly intractable with existing methods for optimal flow control. However, we can attack such problems with the methods that we developed previously and true Newton method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.9 no.4
• /
• pp.420-425
• /
• 1999

The maximum demand controller is an electrical equipment installed at the consumer side of power system for monitoring the electrical energy consumed during every integrating period and preventing the target maximum demand (MD) being exceeded by disconnecting sheddable loads. By avoiding the peak loads and spreading the energy requirement the controller contributes to maximizing the utility factor of the generator systems. It results in not only saving the energy but also reducing the budget for constructing the natural base facilities by keeping thc number of generating plants ~ninimumT. he conventional MD controllers often bring about the large number of control actions during the every inteyating period and/or undesirable loaddisconnecting operations during the beginning stage of the integrating period. These make the users aviod the MD controllers. In this paper. fuzzy control technique is used to get around the disadvantages of the conventional MD control system. The proposed MD controller consists of the predictor module and the fuzzy MD control module. The proposed forecasting method uses the SOFM neural network model, differently from time series analysis, and thus it has inherent advantages of neural network such as parallel processing, generalization and robustness. The MD fuzzy controller determines the sensitivity of control action based on the time closed to the end of the integrating period and the urgency of the load interrupting action along the predicted demand reaching the target. The experimental results show that the proposed method has more accurate forecastinglcontrol performance than the previous methods.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.5A
• /
• pp.649-656
• /
• 2008

The enhancement of the service life of damaged or cracked structures is a major issue for researchers and engineers. The hierarchic void element based on the integrals of Legendre polynomials is used to characterize the fracture behaviour of unpatched crack as well as repaired crack with bonded composite patches by computing the stress intensity factors and stress contours at the crack tip. Since the equivalent single layer approach is adopted in this study, the proposed element is necessary to represent a discontinuous crack part as a continuum body with zero stiffness. Thus the aspect ratio of this element to represent the crack should be extremely slender. The sensitivity of numerical solution with respect to energy release rate, displacement and stress has been tested to show the robustness of zero stiffness element as the aspect ratio is increased up to 2000. The stiffness derivative method and displacement extrapolation method have been applied to calculate the stress intensity factors of Mode I problem. It is noted that the proposed hierarchical void element can be one of alternatives to analyze the patched crack problems.