• Convergence Security Journal
• /
• v.22 no.3
• /
• pp.25-32
• /
• 2022

Recently, the image processing industry has been activated as deep learning-based technology is introduced in the image recognition and detection field. With the development of deep learning technology, learning model vulnerabilities for adversarial attacks continue to be reported. However, studies on countermeasures against poisoning attacks that inject malicious data during learning are insufficient. The conventional countermeasure against poisoning attacks has a limitation in that it is necessary to perform a separate detection and removal operation by examining the training data each time. Therefore, in this paper, we propose a technique for reducing the attack success rate by applying modifications to the training data and inference data without a separate detection and removal process for the poison data. The One-shot kill poison attack, a clean label poison attack proposed in previous studies, was used as an attack model. The attack performance was confirmed by dividing it into a general attacker and an intelligent attacker according to the attacker's attack strategy. According to the experimental results, when the proposed defense mechanism is applied, the attack success rate can be reduced by up to 65% compared to the conventional method.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.14 no.3
• /
• pp.236-256
• /
• 2011

Forest carbon stocks change due to land use change is an important data required by UNFCCC(United Nations framework convention on climate change). Spatially explicit estimation of forest carbon stocks based on IPCC GPG(intergovernmental panel on climate change good practice guidance) tier 3 gives high reliability. But a current estimation which was aggregated from NFI data doesn't have detail forest carbon stocks by polygon or cell. In order to improve an estimation remote sensing and GIS have been used especially in Europe and North America. We divided research trends in main countries into 4 categories such as remote sensing, GIS, geostatistics and environmental modeling considering spatial heterogeneity. The easiest way to apply is combination NFI data with forest type map based on GIS. Considering especially complicated forest structure of Korea, geostatistics is useful to estimate local variation of forest carbon. In addition, fine scale image is good for verification of forest carbon stocks and determination of CDM site. Related domestic researches are still on initial status and forest carbon stocks are mainly estimated using k-nearest neighbor(k-NN). In order to select suitable method for forest in Korea, an applicability of diverse spatial data and algorithm must be considered. Also the comparison between methods is required.

• Journal of Arbitration Studies
• /
• v.24 no.2
• /
• pp.33-52
• /
• 2014

Although climate change is a global scale question, some concerns have been raised that principles of investment arbitration may not adequately address the domestic implementation of climate change measures. A recent ICSID investment arbitration of Vattenfall v. Germany with regard to the investor's alleged damages from the phase-out of nuclear plants is a salient climate change case. The 2005 Kyoto Protocol was made to reduce greenhouse gas emissions and it provides a number of flexible mechanisms such as Joint Implementation (JI) and Clean Development Mechanism (CDM). Implementation of the Kyoto Protocol allows dispute settlement through investor-state arbitration. Any initiation of stricter emission standards can violate the prohibition on expropriations in investment agreements, regardless of the measures created to reduce greenhouse gas emissions. The effect-based expropriation doctrine can charge changes to existing emission standards as interference with the use of property that goes against the legitimate expectation of a foreign investor. In regulatory chill, threat of investor claims against the host state may preclude the strengthening of climate change measures. Stabilization clauses also have a freezing effect on the hosting state's regulation and a new law applicable to the investment. In the fair and equitable standard, basic expectations of investors when entering into earlier carbon-intensive operations can be affected by a regulation seeking to change into a low-carbon approach. As seen in the Methanex tribunal, a non-discriminatory and public purpose of environmental protection measures should be considered as non-expropriation in the arbitral tribunal unless its decision would intentionally impede a foreign investor's investment.