• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Journal of Korean Neurosurgical Society
• /
• v.29 no.12
• /
• pp.1606-1611
• /
• 2000

Objective : To obtain more reliable sample in stereotactic biopsy, authors adopted proton chemical shift imaging ($^1H$-CSI)-directed biopsy. Until now, proton single voxel spectroscopy($^1H$-SVS) technique has been reported as a technique using metabolic information in stereotactic biopsy. The authors performed $^1H$-CSI with a stereotactic headframe in place and evaluated the pathologic results obtained from local metabolic information through $^1H$-CSI. Methods : $^1H$ CSI-directed stereotactic biopsy was performed in four patients. $^1H$-CSI and conventional Gd-enhancement stereotactic MRI was done simultaneously after application of the stereotatic frame. After reconstruction of metabolic maps of NAA/Cr, Cho/Cr, and Lactate/Cr ratios, the focal areas of increased Cho/Cr ratios and decreased NAA/Cr ratios were selected for target sites in the MR images Results : There was no difficulty in performing $^1H$-CSI with the stereotactic headframe in place. In pathologic examinations, the samples taken in area of increased Cho/Cr ratios and decreased NAA/Cr ratios showed the features of increased cellularity, mitoses and cellular atypism, thus facilitated the diagnosis. The pathologic samples taken from the area of increased Lactate/Cr ratios showed prominent feature of necrosis. Conclusion : $^1H$-CSI was feasible with stereotactic head frame in place. The final pathologic results obtained in our samples were concordant with the local metabolic informations from $^1H$-CSI. Authors believe that $^1H$ CSI-directed stereotactic biopsy may provide us advantages in obtaining more reliable tissue specimen in stereotactic biopsy.

• Journal of the Society of Disaster Information
• /
• v.8 no.3
• /
• pp.287-296
• /
• 2012

This study is analysis of the existing domestic terrorism factors and suggest countermeasures for political leadership that can occur to the precautionary terror occasion of the presidential election approached 100 days in the future. Political terror caused mainly achieved peak of Ideology or social conflict by Politically alienated class. Common cause is most likely to occur when the party or the government want to attack a target or can not stand disadvantaged and lose that I suppert the party or the government. 21st century political terror is occurred with Misfits of complaints expressed or social conflict more frequency than attack used random people, new weqpon and bomb. It is interpreted that issued as a drastic measure from individual complaint because of increased polarization of wealth by separate into group and fallen law and order but also we are not familiar with the political compromises. Personal safety escort mission is essential condition of political activity that Absolute personal protection and to come up with a large number of voters for best result. Thus security guard ought to professional security activity depend on gathered situation, location and regional feature for perfect protection service.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.47 no.3
• /
• pp.11-21
• /
• 2010

To aim at improving performance and reflecting user's needs of retrieval, the number of researches has been actively conducted in recent year as the quantity of information and generation of the web pages exceedingly increase. One of alternative approaches can be a tagging system. It makes users be able to provide a representation of metadata including writings, pictures, and movies etc. called tag and be convenient in use of retrieval of internet resources. Tags similar to keywords play a critical role in maintaining target pages. However, they still needs time consuming labors to annotate tags, which sometimes are found to be a hinderance caused by overuse of tagging. In this paper, we present an automatic tagging scheme for a solution of current tagging system conveying drawbacks and inconveniences. To realize the approach, face recognition-based tagging system on SNS is proposed by building a face area detection procedure, linear-based classification and boosting algorithm. The proposed novel approach of tagging service can increase possibilities that utilized SNS more efficiently. Experimental results and performance analysis are shown as well.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.2C
• /
• pp.141-149
• /
• 2007

Associating the shoulder line with head location of the human body is useful in verifying, localizing and tracking persons in an image. Since the head line and the shoulder line, what we call ${\Omega}$-shape, move together in a consistent way within a limited range of deformation, we can build a statistical shape model using Active Shape Model (ASM). However, when the conventional ASM is applied to ${\Omega}$-shape fitting, it is very sensitive to background edges and clutter because it relies only on the local edge or gradient. Even though appearance is a good alternative feature for matching the target object to image, it is difficult to learn the appearance of the ${\Omega}$-shape because of the significant difference between people's skin, hair and clothes, and because appearance does not remain the same throughout the entire video. Therefore, instead of teaming appearance or updating appearance as it changes, we model the discriminative appearance where each pixel is classified into head, torso and background classes, and update the classifier to obtain the appropriate discriminative appearance in the current frame. Accordingly, we make use of two features in fitting ${\Omega}$-shape, edge gradient which is used for localization, and discriminative appearance which contributes to stability of the tracker. The simulation results show that the proposed method is very robust to pose change, occlusion, and illumination change in tracking the head and shoulder line of people. Another advantage is that the proposed method operates in real time.

• 한국정보통신설비학회:학술대회논문집
• /
• 2008.08a
• /
• pp.211-217
• /
• 2008

Recently, there are many efforts to support seamless mobility in 802.11 WLANs using IP Layer mobility protocols. The IP layer mobility protocols are the most efficient mechanism to guarantee the service session continuity when IP subnet is changed during handover. Even if the IP layer mobility protocols are quite efficient, the feature of the protocols that had been designed to consider only L3 layer makes it difficult to improve the performance of hand over more and more. Nowadays, to overcome this limitation of IP mobility protocols, many researchers have worked on the mobility protocols integration of different layers (e.g., L2 layer). In this paper, we propose the enhanced Proxy MIPv4 to minimize the latency of handover using MIH protocol in 802.11 WLANs. The proposed mechanism minimizes the latency of authentication by exchanging security keys between Access Routers during handover. Moreover, it also minimizes packet losses by Inter-AP Tunneling and data forwarding.

• Korean Journal of Remote Sensing
• /
• v.23 no.6
• /
• pp.493-500
• /
• 2007

In general, it accomplished a spectral reflectance analysis to be, the measurement results appear differently by targets, methods and condition. This paper presents a standard methodology for preprocessing mineral/rock samples and setting the distance from a target to the sensor, and then examines closely the spectral features for pyrophyllite. The size of mineral/rock samples is various according to the condition and scale of outcrop, so it is important to maintain the distance between the sensor and the sample. Before standardization for preprocessing samples and the sensor and sample distance, we prepare various rock samples (Quartz Porphyry) such as natural rock, pebble, powder and cutting rock. For a qualitative analysis to minimize the effect of surface condition of the sample and shadow, we maintains the distance from the sample to the sensor at 30cm and measures three times repeatedly for cutting the sample at $1{\sim}2cm$ thickness. To illustrate the proposed methodology, a case study for pyrophyllite was carried out. In this study, pyrophyllite showed an absorption pattern at wave length of 1.406nm, 1,868nm, 2.180nm and 2.309nm, and a higher grade represented strong absorption at 1.406nm and 2.180 nm. These absorption feature corresponds the band 7 of LANDSAT TM and band 8 of ASTER imageries. So, using these results, pyrophyllite deposits were extracted from other features (such as barren area, concrete area, bed of river, stone pit area etc.).

• The Journal of the Korea institute of electronic communication sciences
• /
• v.14 no.5
• /
• pp.907-916
• /
• 2019

In this paper, we propose a new five-qubit multiple bit flip code that can completely protect the target qubit from all multiple bit flip errors using only CNOT gates. The proposed multiple bit flip codes can be easily extended to multiple phase flip codes by embedding Hadamard gate pairs in the root error section as in conventional single bit flip code. The multiple bit flip code and multiple phase flip code in this paper share the state vector error information by four auxiliary qubits. These four-qubit state vectors reflect the characteristic that all the multiple flip errors with Pauli X and Z corrections commonly include a specific root error. Using this feature, this paper shows that low-cost implementation is possible despite the QECC design for multiple-flip error correction by batch processing the detection and correction of Pauli X and Z root errors with only three CNOT gates. The five-qubit multiple bit flip code and multiple phase flip code proposed in this paper have 100% error correction rate and 50% error discrimination rate. All QECCs presented in this paper were verified using QCAD simulator.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.5
• /
• pp.1666-1689
• /
• 2021

The red line of Permanent Basic Farmland is the most important part in the "three-line" demarcation of China's national territorial development plan. The scientific and reasonable delineation of the red line is a major strategic measure being taken by China to improve its ability to safeguard the practical interests of farmers and guarantee national food security. The delineation of Permanent Basic Farmland zoning (DPBFZ) is essentially a multi-objective optimization problem. However, the traditional method of demarcation does not take into account the synergistic development goals of conservation of cultivated land utilization, ecological conservation, or urban expansion. Therefore, this research introduces the idea of artificial immune optimization and proposes a multi-objective model of DPBFZ red line delineation based on a clone selection algorithm. This research proposes an objective functional system consisting of these three sub-objectives: optimal quality of cropland, spatially concentrated distribution, and stability of cropland. It also takes into consideration constraints such as the red line of ecological protection, topography, and space for major development projects. The mathematical formal expressions for the objectives and constraints are given in the paper, and a multi-objective optimal decision model with multiple constraints for the DPBFZ problem is constructed based on the clone selection algorithm. An antibody coding scheme was designed according to the spatial pattern of DPBFZ zoning. In addition, the antibody-antigen affinity function, the clone mechanism, and mutation strategy were constructed and improved to solve the DPBFZ problem with a spatial optimization feature. Finally, Tongxu County in Henan province was selected as the study area, and a controlled experiment was set up according to different target preferences. The results show that the model proposed in this paper is operational in the work of delineating DPBFZ. It not only avoids the adverse effects of subjective factors in the delineation process but also provides multiple scenarios DPBFZ layouts for decision makers by adjusting the weighting of the objective function.

• Journal of the Korean Society of Radiology
• /
• v.17 no.3
• /
• pp.423-431
• /
• 2023

The purpose of this study is to analyze the factors affecting the classification of the severity of contrast media side effects based on the patient's body information using artificial intelligence techniques to be used as basic data to reduce the degree of contrast medium side effects. The data used in this study were 606 examiners who had no contrast medium side effects in the past history survey among 1,235 cases of contrast medium side effects among 58,000 CT scans performed at a general hospital in Seoul. The total data is 606, of which 70% was used as a training set and the remaining 30% was used as a test set for validation. Age, BMI(Body Mass Index), GFR(Glomerular Filtration Rate), BUN(Blood Urea Nitrogen), GGT(Gamma Glutamyl Transgerase), AST(Aspartate Amino Transferase,), and ALT(Alanine Amiono Transferase) features were used as independent variables, and contrast media severity was used as a target variable. AUC(Area under curve), CA(Classification Accuracy), F1, Precision, and Recall were identified through AdaBoost, Tree, Neural network, SVM, and Random foest algorithm. AdaBoost and Random Forest show the highest evaluation index in the classification prediction algorithm. The largest factors in the predictions of all models were GFR, BMI, and GGT. It was found that the difference in the amount of contrast media injected according to renal filtration function and obesity, and the presence or absence of metabolic syndrome affected the severity of contrast medium side effects.