• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1111-1123
• /
• 2023

In recent years, as generative models have developed, research that threatens them has also been actively conducted. We propose a new membership inference attack against text-to-image model. Existing membership inference attacks on Text-to-Image models produced a single image as captions of query images. On the other hand, this paper uses personalized embedding in query images through Textual Inversion. And we propose a membership inference attack that effectively generates multiple images as a method of generating Adversarial Prompt. In addition, the membership inference attack is tested for the first time on the Stable Diffusion model, which is attracting attention among the Text-to-Image models, and achieve an accuracy of up to 1.00.

• Geophysics and Geophysical Exploration
• /
• v.26 no.2
• /
• pp.84-93
• /
• 2023

The supervised learning-based deep-learning seismic inversion techniques have demonstrated successful performance in synthetic data examples targeting small-scale areas. The supervised learning-based deep-learning seismic inversion uses time-domain wavefields as input and subsurface velocity models as output. Because the time-domain wavefields contain various types of wave information, the data size is considerably large. Therefore, research applying supervised learning-based deep-learning seismic inversion trained with a significant amount of field-scale data has not yet been conducted. In this study, we predict subsurface velocity models using Laplace-domain wavefields as input instead of time-domain wavefields to apply a supervised learning-based deep-learning seismic inversion technique to field-scale data. Using Laplace-domain wavefields instead of time-domain wavefields significantly reduces the size of the input data, thereby accelerating the neural network training, although the resolution of the results is reduced. Additionally, a large grid interval can be used to efficiently predict the velocity model of the field data size, and the results obtained can be used as the initial model for subsequent inversions. The neural network is trained using only synthetic data by generating a massive synthetic velocity model and Laplace-domain wavefields of the same size as the field-scale data. In addition, we adopt a towed-streamer acquisition geometry to simulate a marine seismic survey. Testing the trained network on numerical examples using the test data and a benchmark model yielded appropriate background velocity models.

• Atmosphere
• /
• v.15 no.4
• /
• pp.229-239
• /
• 2005

The effect of inversion layer on the land and sea breeze near the Gangneung city was investigated. The land and sea breeze occurrence days were selected, and the height and the intensity of inversion layer were calculated with the upper air observational data of the Sokcho Station. The relationships between the temperature variation near the Gangneung and the inflow time, inland penetration and the inflow depth of the land and sea breeze were also analyzed. And the Gangwon Short-range prediction system was verified with the comparison of surface stream line by the Gangwon short-range prediction system with the AWS wind vector data. It was revealed that the inversion layer tended to block the sea breeze, shorten the inland penetration distance and lower the inflow depth, causing the temperature rise. The comparison and analysis of surface steam line by the Gangwon short-range prediction system and the AWS wind vector showed that the system quite well simulated the sea breeze, thus the system could be well utilized in the prediction of land and sea breeze.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.42 no.3 s.333
• /
• pp.9-16
• /
• 2005

The mobility in strained Si inversion layer on $Si_{1-x}Ge_x$ is calculated considering a quantum effect(subband energy and wavefunction) in inversion layer and relaxation time approximation. The quantum effect in inversion layer is obtained by using self-consistent calculation of $Schr\ddot{o}dinger$ and Poisson equations. For the relaxation time, intravalley and intervalley scatterings are considered. The result shows that the reason for the enhancement in mobility as Ge mole fraction increases is that the electron mobility in 2-폴드 valleys is about 3 times higher than that of 4-폴드 valleys and most electrons are located in 2-폴드 valleys as Ge mole fraction increases. Meanwhile, for the phonon-limited mobility the fitting to experimental data, Coulomb and surface roughness mobilities are included in total mobility, Deformation potentials are selected for the calculated effective field, temperature, and Ge mole fraction dependent mobilities to be fitted to experimental data, and then upgraded data can be obtained by considering nonparabolicity in Si band structure.

• Geophysics and Geophysical Exploration
• /
• v.1 no.1
• /
• pp.71-78
• /
• 1998

Inversion schemes of 2-D MT survey data generally take enormous computational time and computer memory. In addition, careful attention must be paid in handling MT data, especially in cases of TM mode, inversion results can be seriously distorted because of static effect caused by current channeling across inhomogeneous surface boundaries. There-fore inversion algorithm using the GRRI scheme for TM mode MT data was implemented. This scheme is based on a perturbation analysis with a locally 2-D analysis and local inversions were sequently performed over each divided section without additional forward modelings. The algorithm was applied to several synthetic data for the purpose of verification of its efficiency and applicability. With less computer resources than conventional schemes, it could handle static effect directly by including current channeling across inhomogeneous boundaries. Thus it is expected to be used for an useful tool such as a real-time inversion scheme in the field.

• Journal of the Korean Society for Precision Engineering
• /
• v.21 no.11
• /
• pp.130-139
• /
• 2004

The Casing Oscillator is a bore file Equipment for the all-casing process. All-casing process is a method of foundation work in construction yard to oscillate steel Casing in the ground. The existing Casing Oscillator has some problem like not boring horizontally with disturbance and not driving Casing othor angle except horizon. To solve problem, the new structure Casing Oscillator is presented and studied. The performance of Casing Oscillator is improved by kinematics analysis. The Casing Oscillator is similar to the parallel manipulator in structure. So we obtain Inverse kinematics solution of Casing Oscillator easily. But it is difficult to solve forward kinematics of Casing Oscillator. T his paper presents a novel pose description corresponding to the structure characteristics of parallel manipulators. Through analysis on geometry theory, we obtain a new method of the closed-form solution to the forward kinematics using Kinematic Inversion. The closed-form solution contains two different meanings -analytical and real-time. So we reach the goal of practical application and control. Closed-form forward kinematics solution is verified by an inverse kinematics analysis. It shows that the method has a practical value for real -time control and inverse kinematics servo control.

• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.7
• /
• pp.658-668
• /
• 2004

This paper is concerned with a control allocation strategy using the dynamic inversion and the pseudo inverse control which generates the nominal control input trajectories. In addition, an autopilot design method is proposed by using time-varying control technique which is time-varying version of the pole placement of linear time-invariant system for an agile missile with aerodynamic fin and thrust vectoring control. The control allocation proposed in this paper is capable of extracting the maximum performance by combining each control effector, aerodynamic fin and thrust vectoring control. The adopted time-varying control technique for the autopilot design enhances the robustness of the tracking performance for a reference command. The main results are validated through the nonlinear simulations with aerodynamic data.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.6 no.3
• /
• pp.122-130
• /
• 2003

This paper is concerned with a control allocation strategy using the dynamic inversion and the pseudo inverse control which generates the nominal control input trajectories, and autopilot design using time-varying control technique which is time-varying version of pole placement of linear time-invariant system for an agile missile with aerodynamic fin and thrust vectoring control. Control allocation of this paper is capable of extracting the maximum performance from each control effector, aerodynamic fin and thrust vectoring control, by combining the action of them. Time-varying control technique for autopilot design enhance the robustness of the tracking performance for a reference command. The main results are validated through the nonlinear simulation.

• 한국지구물리탐사학회:학술대회논문집
• /
• 2006.06a
• /
• pp.117-122
• /
• 2006

In the geophysical monitoring to understand the change of subsurface material properties with time, the time-invariant static subsurface model is commonly adopted to reconstruct a time-lapse image. This assumption of static model, however, can be invalid particularly when fluid migrates very quickly in highly permeable medium in the brine injection experiment. In such case, the resultant subsurface images may be severely distorted. In order to alleviate this problem, we develop a new least-squares inversion algorithm under the assumption that the subsurface model will change continuously in time. Instead of sampling a time-space model into numerous space models with a regular time interval, a few reference models in space domain at different times pre-selected are used to describe the subsurface structure continuously changing in time; the material property at a certain space coordinate are assumed to change linearly in time. Consequently, finding a space-time model can be simplified into obtaining several reference space models. In order to stabilize iterative inversion and to calculate meaningful subsurface images varying with time, the regularization along time axis is introduced assuming that the subsurface model will not change significantly during the data acquisition. The performance of the proposed algorithm is demonstrated by the numerical experiments using the synthetic data of crosshole dc resistivity tomography.

• The Journal of the Acoustical Society of Korea
• /
• v.32 no.4
• /
• pp.308-316
• /
• 2013

A set of experiments (Shallow Water 2006, SW06) was carried out in shallow water near the New Jersey shelf break in summer 2006. Significant fluctuations in direct and surface reflected arrivals were observed from the chirp data (1100~2900 Hz) measured on a vertical line array. This paper presents a geoacoustic inverssion technique for short-range acoustic data with fluctuating arrivals and inversion results of experimental data. In order to reduce effects of random sea surface on the inversion, the acoustic energy back-propagated from the array to the source through direct and bottom-reflected paths is defined as the objective function. A multi-step inversion scheme is applied to the data using VFSR (Very Fast Simulated Reannealing) optimization technique. The inversion results show a source depth oscillation period equal to the measured ocean surface wave period. The inverted bottom sound speed is 1645 m/s and is similar to that estimated by other work at the same site.