• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.5
• /
• pp.895-907
• /
• 2020

With the recent increasing influence of multimedia content other than the text-based content, services that help to process information in content brings us great convenience. These services' representative features are searching and masking the sensitive data. It is not difficult to find the solutions that provide searching and masking function for text information and image. However, even though we recognize the necessity of the technology for searching and masking a part of the audio data, it is not easy to find the solution because of the difficulty of the technology. In this study, we propose web application that provides searching and masking functions for audio data using audio partitioning method. While we are achieving the research goal, we evaluated several speech to text conversion APIs to choose a proper API for our purpose and developed regular expressions for searching sensitive information. Lastly we evaluated the accuracy of the developed searching and masking feature. The contribution of this work is in design and implementation of searching and masking a sensitive information from the audio data by the various functionality proving experiments.

• 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.

• Asian-Australasian Journal of Animal Sciences
• /
• v.21 no.8
• /
• pp.1103-1108
• /
• 2008

An investigation to determine frequency and distribution of folate sensitive chromosomal fragile sites was carried out in a Pakistani breed of Lohi sheep to uncover fragile site phenomena. The means and standard errors of aberrant cell count (AC) and Number of aberrations (NoA) in Lohi sheep were $0.56{\pm}0.15$ and $0.59{\pm}0.16$ in the control cultures. FUdR treated cells showed significantly higher (p<0.001) AC and NoA means ($2.18{\pm}0.33$ and $2.65{\pm}0.50$). The sex comparison for the frequency of expression indicated that males had significantly higher number of aberrant cells and total number of aberrations in FUdR cultures than the female group in Lohi sheep. The comparison of control cultures was however, not significantly different between the two groups. The regression analysis of FUdR-induced chromosomal fragility data analysis of the fragility data predicted very low ${\beta}$ of 0.325 and 0.412 for AC and NoA respectively. Lohi chromosomes expressed lesions in only 7 and 24 bands in the control and FUdR cultures respectively. The total number of significantly fragile bands in the Lohi genome was only 4. The X-chromosome of the Lohi sheep was highly stable at $5{\mu}g/ml$ FUdR with no fragile sites. The sex comparison for the distribution of fragile sites across the Lohi genome did not reveal any noticeable differences.

• Structural Monitoring and Maintenance
• /
• v.5 no.2
• /
• pp.231-242
• /
• 2018

Subway tunnel structure has been rapidly developed in many cities for its strong transport capacity. The model-based damage detection of subway tunnel structure is usually difficult due to the complex modeling of soil-structure interaction, the indetermination of boundary and so on. This paper proposes a new data-based method for the damage detection of subway tunnel structure. The root mean square acceleration and cross correlation function are used to derive a statistical pattern recognition algorithm for damage detection. A damage sensitive feature is proposed based on the root mean square deviations of the cross correlation functions. X-bar control charts are utilized to monitor the variation of the damage sensitive features before and after damage. The proposed algorithm is validated by the experiment of a full-scale two-rings subway tunnel lining, and damages are simulated by loosening the connection bolts of the rings. The results verify that root mean square deviation is sensitive to bolt loosening in the tunnel lining and X-bar control charts are feasible to be used in damage detection. The proposed data-based damage detection method is applicable to the online structural health monitoring system of subway tunnel lining.

• Fashion & Textile Research Journal
• /
• v.19 no.1
• /
• pp.90-100
• /
• 2017

In this research, consumer awareness investigation and wearing test were carried out for obtaining useful data on the development of thermo-sensitive functional clothing material. A survey involved 216 people in Seoul and Kyeonggi-do, and 200 questionnaires data were analyzed by descriptive statistics and frequency using SPSS 17.0. Four healthy men in twenties were participated for wearing test. Subjects in normal loungewear were exposed to temperature change from the initial temperature $30^{\circ}C$ down to $5^{\circ}C$ for an hour in a climate chamber. The environmental temperature, surface temperature of garment and skin were measured. As a result, most of respondents have all season clothing products such as underwear, hosiery, and jogging suit for loungewear. Also, thermo regulator y functional clothes are frequently used as underwear and sweat shirt. The consumer awareness investigation on thermo regulatory functional clothing showed that the most important key buying factor is quick climate temperature response, easy maintenance, design and cost, in that order. Surface temperature of garment went down with the cooling down of environmental temperature. The lower environmental temperature, the greater temperature difference by body part showed. Skin temperature change by environmental temperature showed similar tendency of garment surface temperature. In comparison between garment surface and body skin, temperature difference became larger under the lower environmental temperature.

• Journal of information and communication convergence engineering
• /
• v.10 no.1
• /
• pp.40-44
• /
• 2012

Data mining, which is attracting public attention, is a process of drawing out knowledge from a large mass of data. The key technique in data mining is the ability to maximize the similarity in a group and minimize the similarity between groups. Since grouping in data mining deals with a large mass of data, it lessens the amount of time spent with the source data, and grouping techniques that shrink the quantity of the data form to which the algorithm is subjected are actively used. The current grouping algorithm is highly sensitive to static and reacts to local minima. The number of groups has to be stated depending on the initialization value. In this paper we propose a gene algorithm that automatically decides on the number of grouping algorithms. We will try to find the optimal group of the fittest function, and finally apply it to a data mining problem that deals with a large mass of data.

• Journal of Korea Multimedia Society
• /
• v.20 no.11
• /
• pp.1785-1792
• /
• 2017

Generally, big data contains sensitive information about individuals, and thus directly releasing it for public use may violate existing privacy requirements. Therefore, privacy-preserving data publishing (PPDP) has been actively researched to share big data containing personal information for public use, while protecting the privacy of individuals with minimal data modification. Recently, with increasing demand for big data sharing in various area, there is also a growing interest in the development of software which supports a privacy-preserving data publishing. Thus, in this paper, we develops the system which aims to effectively and efficiently support privacy-preserving data publishing. In particular, the system developed in this paper enables data owners to select the appropriate anonymization level by providing them the information loss matrix. Furthermore, the developed system is able to achieve a high performance in data anonymization by using distributed Hadoop clusters.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.1258-1260
• /
• 2003

Cloud's properties can be showed on different spectral channel. The 0.65${\mu}$m reflectance is mainly function of cloud optical thickness and reflectance of 1.6${\mu}$m is sensitive to cloud phase and particle size distribution. So we can use multi-spectral information to analysis cloud's microphysical properties.

• Journal of Photoscience
• /
• v.3 no.2
• /
• pp.65-69
• /
• 1996

In this paper, it was studied about the red-sensitive layer. UV-Vis data of the dye at monomeric and J-state were considered with respect to their performance(contrast, density and fog) in photographic emulsion. The sensitizing effect of the dye is found to be strongly structuredependent. Naphthothiazolo carbocyanine dye can be used as red-sensitizing dye for the spectral sensitization of photographic emulsion.

• Journal of KIISE
• /
• v.41 no.12
• /
• pp.1081-1089
• /
• 2014

Context-sensitive spelling-error correction methods are largely classified into rule-based methods and statistical data-based methods, the latter of which is often preferred in research. Statistical error correction methods consider context-sensitive spelling error problems as word-sense disambiguation problems. The method divides a vocabulary pair, for correction, which consists of a correction target vocabulary and a replacement candidate vocabulary, according to the context. The present paper proposes a method that integrates a word-phrase n-gram model into a conventional model in order to improve the performance of the probability model by using a correction vocabulary pair, which was a result of a previous study performed by this research team. The integrated model suggested in this paper includes a method used to interpolate the probability of a sentence calculated through each model and a method used to apply the models, when both methods are sequentially applied. Both aforementioned types of integrated models exhibit relatively high accuracy and reproducibility when compared to conventional models or to a model that uses only an n-gram.