• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.3
• /
• pp.1100-1118
• /
• 2021

In a wide range of ML applications, the training data contains privacy-sensitive information that should be kept secure. Training the ML systems by privacy-sensitive data makes the ML model inherent to the data. As the structure of the model has been fine-tuned by training data, the model can be abused for accessing the data by the estimation in a reverse process called model inversion attack (MIA). Although, MIA has been applied to shallow neural network models of recognizers in literature and its threat in privacy violation has been approved, in the case of a deep learning (DL) model, its efficiency was under question. It was due to the complexity of a DL model structure, big number of DL model parameters, the huge size of training data, big number of registered users to a DL model and thereof big number of class labels. This research work first analyses the possibility of MIA on a deep learning model of a recognition system, namely a face recognizer. Second, despite the conventional MIA under the white box scenario of having partial access to the users' non-sensitive information in addition to the model structure, the MIA is implemented on a deep face recognition system by just having the model structure and parameters but not any user information. In this aspect, it is under a semi-white box scenario or in other words a gray-box scenario. The experimental results in targeting five registered users of a CNN-based face recognition system approve the possibility of regeneration of users' face images even for a deep model by MIA under a gray box scenario. Although, for some images the evaluation recognition score is low and the generated images are not easily recognizable, but for some other images the score is high and facial features of the targeted identities are observable. The objective and subjective evaluations demonstrate that privacy cyber-attack by MIA on a deep recognition system not only is feasible but also is a serious threat with increasing alert state in the future as there is considerable potential for integration more advanced ML techniques to MIA.

• Journal of the Korean Society of Environmental Restoration Technology
• /
• v.26 no.6
• /
• pp.33-50
• /
• 2023

This study is to propose the ecosystem service valuation method as a complementary or alternative tool to overcome the limitations of the contingent valuation method(CVM), typically used to assess social benefits in preliminary feasibility studies. With an increasing interest in natural and environmental restoration projects, we assessed social benefits with theses CVM and ecosystem service valuation method from a case of Janghang wetland restoration project and compared the extent of the two social benefits. For quantitative evaluation of ecosystem services, the biophysical quantity for each ecosystem service indicator was calculated and then converted into currency (KRW) units to estimate the economic value of ecosystem services. The four ecosystem regulating service indicators were selected including greenhouse gas capture/storage, air pollution, water quantity and quality regulation. The amounts of CO₂ sequestration and storage as a ecosystem's greenhouse gas regulating service in the study area were 73.04 tCO₂/yr and 5,867.53 tCO₂/yr respectively. The reduction of SO₂, one of air pollutant gases by ecosystems was calculated to be 180.27 kg/yr, the reduction of NO₂ to be 378.90 kg/yr, and the reduction of fine dust (PM10) to be 9,713.92 kg/yr. The amount of freshwater regulating service by the ecosystem was estimated to be 459,394,319ℓ/yr, and the amount of nitrogen in freshwater removed by the ecosystem was 78.00kg/yr. Study results show that the benefits derived from the CVM were KRW 227.8 billion over the 30-year analysis period and those from the ecosystem service valuation method were KRW 41.4 billion for regulatory services and KRW 148.8 billion for cultural services, totaling KRW 189.5 billion. With KRW 184.8 billion of the total costs, the benefit/cost ratio using the CVM was 1.23 and that with the ecosystem service valuation method was 1.03. This study implications include that the CVM and ecosystem service valuation method can be applied together to assess and compare social benefits for natural and environmental restoration projects.

• Korean journal of applied entomology
• /
• v.15 no.1 s.26
• /
• pp.7-16
• /
• 1976

In order to increase utility efficiency of red wood ants, Formica rufa truncicola var. yessonesis Forel as a resource of natural enemy of pine caterpillar, Dendrolimus spectabilis Butler, by finding out ecological and environmental factors in the habitat of red wood ants, the nest distribution and its density in habitat, plant distribution and density, stand-density of red pine, nest building and fixing plants, relative humidity of surface soil, physical and chemical natures of soil, and breeding rate were examined. The obtained results are summarized as follows: 1. The nest of red wood ants was densely distributed, in the lower-and middle top of mountain but no nest was found in the top. 2. The economical distribution of nest of habitat was estimated as $2.85/m^2$ and the lowest density as $1.93/m^2$ and these estimation lead us to confirm that pine caterpillar could be controlled. 3. The ecological characteristics of habitat seemed to be represented as higher stand-density of red pine of 10-20 years of age with large areas of eroded land under trees. The major grasses prevailing in this area were Andropogon brevifolius. Arundinella hirta, Miscanthus purpurasens, Eulia speciosa, Themeda japonica, Cymbopogon goeringii, and Eccoilpus cotulifer 4. Red wood ants seemed to build the nest by using red pine, Arundinella hirta, Miscanthus purpurascens, Themeda japonica or Cymbopogon goeringii as a fixing plant. 5. The limited point of humidity percent in habitat of red wood ants was estimated as $76\%$ during the acting period of May to September and as $72\%$ during pre-period of hibernation of October to November. 6. Soil analysis in habitating region showed higher concentration of organic matters and lower concentration of calcium and magnesium, and habitat was largely composed of silt and fine sand rather than coarse sand. 7. When the separated colony was transplanted to non-habitating red pine forest that seemed to have the similiar conditions as those of habitat, propagation and establishment of nest was possible.

• Journal of Korean Society of Forest Science
• /
• v.98 no.6
• /
• pp.791-798
• /
• 2009

The role of forest and soil carbon under global climate change is getting important as a carbon sink and it is necessary to research on applicable forest models as well as in the field for a study of these dynamics. On this study, historical annual litter dataset as a major input data for the forest soil carbon model, Yasso was established using a dendrochronological reconstruction method, and the soil carbon dynamics of a Pinus densiflora forest in Gwangneung, Korea was simulated using Yasso. The amount of litter (needle, branch, stem and fine root) production, which was estimated using the dendrochronological method, has increased continuously from 1971 to 2006. Furthermore, there was no significant error between estimated and measured values of litter production (needle and branch) in 2006. The average of simulated soil carbon stock up to 30 cm depth was $46.30{\pm}4.28tCha^{-1}$, which accounted for 53% of carbon stock in trees of the forest, and had no significant difference and error with measured soil carbon stock. Under the climate change trend in Korea according to IPCC A1B scenario, it was estimated that the simulated soil carbon stock in the region would increase continuously from 1971 to 2041 and then decreased until 2100. Compared to the result of the scenario that there is no climate change, the soil carbon stock could be decreased up to 7.58% at 2100. It was inferred the dendrochronological reconstruction method and simulation of Yasso model are useful to estimate soil carbon dynamics of the natural P. densiflora forest. Follow-up researches, such as improvement of the dendrochronological method and Yasso model and their application and validation in various environment, are needed to produce more reliable results.

• Korean Journal of Soil Science and Fertilizer
• /
• v.42 no.2
• /
• pp.132-138
• /
• 2009

Soils originated from limestone, located at the southern part of Kangwon province and Jecheon, Danyang of Chungbuk province are mainly composed of fine texture, have different properties from soils originated from granite and granite gneiss, especially for water movement. This study was conducted for making PTF(Pedo-Transfer Function) for Kfs(field saturaton hydraulic conductivity) estimation, and for investigating the relation between soil particle distribution and the infiltration and percolation rate in soils originated from limestone. Soils used for the experiment were 6 soils of Gwarim, Mosan, Jangseong, Maji, Anmi and Pyongan series. Infiltration and percolation rate for the soil were measured by a disc tension infiltrometer and a Guelph permeameter, respectively. The particle size distribution and organic matter content of the soils were analyzed. Kfs was not related with sand, silt, clay, and organic mattrer (OM) content because of forest soils which contained high gravel, pebble, and cobble content, and O layer with high OM content. After Mosan soil series and O layer of Gwarim series were excluded for the data analysis, Kfs was explained as a linear function with sand and clay content and a exponential function with OM content. As a result, the PTF equation was obtained as Kfs=-4.20558+0.479706*(S)+0.023187*exp(1.829*OM) ($R^2=0.6558^{*}$).

• Journal of Radiation Protection and Research
• /
• v.40 no.4
• /
• pp.194-201
• /
• 2015

We have investigated the EPR signal properties in 12 components of two mobile phones (LCD, OLED) using electron paramagnetic resonance (EPR) spectrometer in this study.EPR measurements were performed at normal atmospheric conditions using Bruker EXEXSYS-II E500 spectrometer with X-band bridge, and samples were irradiated by $^{137}Cs$ gamma-ray source. To identify the presence of radiation-induced signal (RIS), the EPR spectra of each sample were measured unirradiated and irradiated at 50 Gy. Then, dose-response curve and signal intensity variating by time after irradiation were measured. As a result, the signal intensity increased after irradiation in all samples except the USIM plastic and IC chip. Among the samples, cover glass(CG), lens, light guide plate(LGP) and diffusion sheet have shown fine linearity ($R^2$ > 0.99). Especially, the LGP had ideal characteristics for dosimetry because there were no signal in 0 Gy and high rate of increase in RIS. However, this sample showed weakness in fading. Signal intensity of LGP and Diffusion Sheet decreased by 50% within 72 hours after irradiation, while signals of Cover Glass and Lens were stably preserved during the short period of time. In order to apply rapidly EPR dosimetry using mobile phone components in large-scale radiation accidents, further studies on signal differences for same components of the different mobile phone, fading, pretreatment of samples and processing of background signal are needed. However, it will be possible to do dosimetry by dose-additive method or comparative method using unirradiated same product in small-scale accident.

• Korean Journal of Soil Science and Fertilizer
• /
• v.43 no.1
• /
• pp.1-7
• /
• 2010

This study was carried out to investigate plow pan characteristics and to grasp the relationship between its particle size distribution fractal dimension ($D_m$) and water transport in paddy plow pan. Twenty four soil sampling sites with different management groups, ordinary and sandy-textured, were selected and investigated for physical properties of soils such as Yamanaka hardness in April, non-submerged condition, before rice seedling transplanting. The plow pan appearing depth and thickness was determined by penetration resistance profile. Undisturbed core samples with five replicates were sampled at plow pan layerwith 2 inch cores for measuring soil bulk density and saturated hydraulic conductivity. The particle size distribution fractal dimension ($D_m$) was calculated by the method following the procedure Tylerand Wheacraft (1992), using the USDA-based particle size analysis datawith fractions of 0-0.002, 0.002-0.053, 0.053-0.1, 0.1-0.25, 0.25-0.5, 0.5-1.0, and 1.0-2.0 mm. The plow pan of investigated fields appeared at a range from 5 to 30 cm depth, showing minimum value in sandy-textured management group and maximum value in ordinary management group. The thickness of plow pan were distributed from 5 to 17 cm, showing both minimum and maximum values in sandy-textured management group. Averagely, the plow appearing depth were deeper in ordinary management group than in sandy-textured management group, whereas the reverse in the thickness of plow pan. The particle size distribution fractal dimension ($D_m$) had higher value with finer textures, with higher fractality in coarser texture. Saturated hydraulic conductivities, $K_s$, of plow pan soils distributed from 0.5 to 1420 mm $day^{-1}$, having the highest value in sandy skeletal soils. The $K_s$ decreased with decreasing clay content and $D_m$, showing power function relationships. The coefficient of determination, $R^2$, of the fitted power functions were higher in $D_m$ as x-axis than in clay content. This means that $D_m$ could give us more effective estimation than clay content. Especially, sandy-textured paddy soils had higher $R^2$, compared to ordinary paddy soils. $K_s$ of relatively coarse-textured soils with less than 18%of clay content, therefore, was more dependent on particle size distribution than that of relatively fine-textured soils. From these results, it could be concluded that the fractal scaling gives us a unique quantity describing particle size distribution and then can be applied to estimate saturated hydraulic conductivity, especially more effective in coarse-textured soils.

• Korean Journal of Remote Sensing
• /
• v.37 no.2
• /
• pp.275-290
• /
• 2021

Sulfur dioxide (SO2) in the atmosphere is mainly generated from anthropogenic emission sources. It forms ultra-fine particulate matter through chemical reaction and has harmful effect on both the environment and human health. In particular, ground-level SO2 concentrations are closely related to human activities. Satellite observations such as TROPOMI (TROPOspheric Monitoring Instrument)-derived column density data can provide spatially continuous monitoring of ground-level SO2 concentrations. This study aims to propose a 2-step residual corrected model to estimate ground-level SO2 concentrations through the synergistic use of satellite data and numerical model output. Random forest machine learning was adopted in the 2-step residual corrected model. The proposed model was evaluated through three cross-validations (i.e., random, spatial and temporal). The results showed that the model produced slopes of 1.14-1.25, R values of 0.55-0.65, and relative root-mean-square-error of 58-63%, which were improved by 10% for slopes and 3% for R and rRMSE when compared to the model without residual correction. The model performance by country was slightly reduced in Japan, often resulting in overestimation, where the sample size was small, and the concentration level was relatively low. The spatial and temporal distributions of SO2 produced by the model agreed with those of the in-situ measurements, especially over Yangtze River Delta in China and Seoul Metropolitan Area in South Korea, which are highly dependent on the characteristics of anthropogenic emission sources. The model proposed in this study can be used for long-term monitoring of ground-level SO2 concentrations on both the spatial and temporal domains.

• Journal of Korea Water Resources Association
• /
• v.55 no.12
• /
• pp.1041-1052
• /
• 2022

This study is to estimate the optimal ecological flow and analysis the spatial distribution of fish habitat for Andong dam downstream reach (4,565.7 km²) using PHABSIM (Physical Habiat Simulation System) and River2D. To establish habitat models, the cross-section informations and hydraulic input data were collected uisng the Nakdong river basic plan report. The establishment range of PHABSIM was set up about 410.0 m from Gudam streamflow gauging station (GD) and about 6.0 km including GD for River2D. To select representative fish species and construct HSI (Habitat Suitability Index), the fish survey was performed at Pungji bridge where showed well the physical characteristics of target stream located downstream of GD. As a result of the fish survey, Zacco platypus was showed highly relative abundance resulting in selecting as the representative fish species, and HSI was constructed using physical habitat characteristics of the Zacco platypus. The optimal range of HSI was 0.3~0.5 m/s at the velocity suitability index, 0.4~0.6 m at the depth suitability index, and the substrate was sand to fine gravel. As a result of estimating the optimal ecological flow by applying HSI to PHABSIM, the optimal ecological flow for target stream was 20.0 m³/sec. As a result of analysis two-dimensional spatial analysis of fish habitat using River2D, WUA (Weighted Usable Area) was estimated 107,392.0 m²/1000 m under the ecological flow condition and it showed the fish habitat was secured throughout the target stream compared with Q₃₅₅ condition.