• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.2
• /
• pp.632-657
• /
• 2022

Detecting web attacks is a major challenge, and it is observed that the use of simple models leads to low sensitivity or high false positive problems. In this study, we aim to develop a robust two-stage deep learning based stacked ensemble web application firewall. Normal and abnormal classification is carried out in the first stage of the proposed WAF model. The classification process of the types of abnormal traffics is postponed to the second stage and carried out using an integrated stacked ensemble model. By this way, clients' requests can be served without time delay, and attack types can be detected with high sensitivity. In addition to the high accuracy of the proposed model, by using the statistical similarity and diversity analyses in the study, high generalization for the ensemble model is achieved. Within the study, a comprehensive, up-to-date, and robust multi-class web anomaly dataset named GAZI-HTTP is created in accordance with the real-world situations. The performance of the proposed WAF model is compared to state-of-the-art deep learning models and previous studies using the benchmark dataset. The proposed two-stage model achieved multi-class detection rates of 97.43% and 94.77% for GAZI-HTTP and ECML-PKDD, respectively.

• Membrane Journal
• /
• v.27 no.5
• /
• pp.449-457
• /
• 2017

In this study, the 10 stages stacked module was designed by increasing the number of unit cells in the membrane capacitive deionization(MCDI) process. The aminated polysulfone and sulfonated poly(ether ether ketone) were synthesized and coated on porous carbon electrode by casting method. The salt removal efficiency was measured for the 10 stage stacked module under the operation conditions of adsorption voltage and time, desorption voltage and time, flow rate and concentration of feed water, and di-valent solutions including $CaSO_4$, $MgCl_2$ and tap water. Typically, when 100 mg/L of NaCl as the feed was used, the salt removal efficiency was 98.3% at a flow rate of 100 mL/min, the adsorption condition of 1.2 V/3 min and desorption condition of -0.5 V/5 min.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2004.10a
• /
• pp.304-307
• /
• 2004

As optical fiber communication grows, the fiber alignment become the focus of industrial attention. This greatly influence the overall production rates for the opto-electric products. We proposed multi-axis nano positioning stage for optical fiber alignment. This device has 3 DOF translation and sub nanometer resolution. This nano stage consist of 3 PZT-driven flexure stages which are stacked parallel. The displacement of it is measured with capacitance gauge and is controlled by computer-embedded main controller. The design process of flexure stage using FEM is proposed and the performance evaluation of this system is verified with experiments.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 2004.10a
• /
• pp.213-218
• /
• 2004

As optical fiber communication grows, the fiber alignment become the focus of industrial attention. This greatly influence the overall production rates for the opto-electric products. We proposed multi-axis nano positioning stage for optical fiber alignment. This device has 3 DOF translation and sub nanometer resolution. This nano stage consist of 3 PZT-driven flexure stages which are stacked parallel. The displacement of it is measured with capacitance gauge and is controlled by computer-embedded main controller. The design process of flexure stage using FEM is proposed and the performance evaluation of this system is verified with experiments.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.

• Journal of the Korean Society of Industry Convergence
• /
• v.22 no.5
• /
• pp.519-527
• /
• 2019

This study was conducted to present useful information on the utilization and productivity analysis of laminated modular multi-family housing. To this end, the process of site installation was investigated and analyzed through a prior study, and the monitoring survey was conducted through the site installation case of an stacked multi-family housing. Based on the above, the results of productivity analysis using the web-cyclone technique were also presented. However, the site installation process has limitations on generalisation because the process is not the same for each construction company, and also limits that require verification through application in the actual site of the web-cyclone model presented in this study.

• Macromolecular Research
• /
• v.12 no.1
• /
• pp.1-10
• /
• 2004

The structural changes occurring in the isothermal crystallization processes of polyethylene (PE), poly-oxymethylene (POM), and vinylidene fluoridetrifluoroethylene (VDFTrFE) copolymer have been reviewed on the basis of our recent experimental data collected by the time-resolved measurements of synchrotron-sourced wide-angle (WAXS) and small-angle X-ray scatterings (SAXS) and infrared spectra. The temperature jump from the melt to a crystallization temperature could be measured at a cooling rate of 600-1,000 $^{\circ}C$/min, during which we collected the WAXS, SAXS, and infrared spectral data successfully at time intervals of ca. 10 sec. In the case of PE, the infrared spectral data clarified the generation of chain segments of partially disordered trans conformations immediately after the jump. These segments then became transformed into more-regular all-trans-zigzag forms, followed by the formation of an orthorhombic crystal lattice. At this stage, the generation of a stacked lamella structure having an 800-${\AA}$-long period was detected in the SAXS data. This structure was found to transfer successively to a more densely packed lamella structure having a 400-${\AA}$-long period as a result of the secondary crystallization of the amorphous region in-between the original lamellae. As for POM, the formation process of a stacked lamella structure was essentially the same as that mentioned above for PE, as evidenced from the analysis of SAXS and WAXS data. The observation of morphology-sensitive infrared bands revealed the evolution of fully extended helical chains after the generation of lamella having folded chain structures. We speculate that these extended chains exist as taut tie chains passing continuously through the neighboring lamellae. In the isothermal crystallization of VDFTrFE copolymer from the melt, a paraelectric high-temperature phase was detected at first and then it transferred into the ferroelectric low-temperature phase at a later stage. By analyzing the reflection profile of the WAXS data, the structural ordering in the high-temperature phase and the ferroelectric phase transition to the low-temperature phase of the multi-domain structure were traced successfully.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.16 no.2 s.93
• /
• pp.154-160
• /
• 2005

Microwave Otters are essential device in modem wireless systems. A compact dimension BPF(Band-pass Filter) for IEEE 802.11a WLAN service is realized using LTCC multi-layer process. To extrude 2-stage band-pass equivalent circuit, band-pass and J-inverter transform applied to Chebyshev low-pass prototype filter. Because parallel L-C resonator is complicate and hard to control the inductor characteristics in high frequency, the shorted $\lambda/4$ stripline is selected for the resonator structure. The passive element is located in the different layers connected by conventional via structure and isolated by inner GND. The dimension of fabricated stacked band-pass filter which is composed of six layers, is $2.51\times2.27\times1.02\;mm^3$. The measured filter characteristics show the insertion loss of -2.25 dB, half-power bandwidth of 220 MHz, attenuation at 5.7 GHz of -32.25 dB and group delay of 0.9 ns at 5.25 GHz.

• Journal of Energy Engineering
• /
• v.20 no.3
• /
• pp.185-190
• /
• 2011

This paper describes design, fabrication, and evaluation of the conduction cooled high temperature superconducting (HTS) magnet for superconducting magnetic energy storage (SMES). The HTS magnet is composed of twenty-two of double pancake coils made of 4-ply conductors that stacked two Bi-2223 multi-filamentary tapes with the reinforced brass tape. Each double pancake coil consists of two solenoid coils with an inner diameter of 500 mm, an outer diameter of 691 mm, and a height of 10 mm. The aluminum plates of 3 mm thickness were arranged between double pancake coils for the cooling of the heat due to the power dissipation in the coil. The magnet was cooled down to 5.6 K with two stage Gifford McMahon (GM) cryocoolers. The maximum temperature at the HTS magnet in discharging mode rose as the charging current increased. 1 MJ of magnetic energy was successfully stored in the HTS magnet when the charging current reached 360A without quench. In this paper, thermal and electromagnetic behaviors on the conduction cooled HTS magnet for SMES are presented and these results will be utilized in the optimal design and the stability evaluation for conduction cooled HTS magnets.