• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.5
• /
• pp.169-176
• /
• 2020

Recently, deep neural networks (DNNs) have brought revolutions to many mobile and embedded devices by providing human-level machine intelligence for various applications. However, high inference accuracy of such DNNs comes at high computational costs, and, hence, there have been significant efforts to reduce computational overheads of DNNs either by compressing off-the-shelf models or by designing a new small footprint DNN architecture tailored to resource constrained devices. One notable recent paradigm in designing small footprint DNN models is sharing parameters in several layers. However, in previous approaches, the parameter-sharing techniques have been applied to large deep networks, such as ResNet, that are known to have high redundancy. In this paper, we propose a parameter-sharing method for already parameter-efficient small networks such as ShuffleNetV2. In our approach, small templates are combined with small layer-specific parameters to generate weights. Our experiment results on ImageNet and CIFAR100 datasets show that our approach can reduce the size of parameters by 15%-35% of ShuffleNetV2 while achieving smaller drops in accuracies compared to previous parameter-sharing and pruning approaches. We further show that the proposed approach is efficient in terms of latency and energy consumption on modern embedded devices.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.15 no.5
• /
• pp.2641-2654
• /
• 2014

In the smart media environment, magazine industry has been experiencing a transition to ecosystem of value network, which includes high complexity and ambiguity. Using case study method, this article conducts research on digital convergence, the model of magazine ecosystem and adaptation strategy of global magazine companies. Research findings have it that the way of contents production of global magazines has been based on collaborative production system within communities, expert communities, creative users, media contents companies and magazine platform. The system shows different patterns and characteristics depending on magazine-driven platform, Platform-driven platform or user-driven platform. Collaboration system has been confirmed in various cases: Huffington Post and Zinio which collaborate with media contents companies, Amazon magazines and Bookish with magazine companies, Huffington Post and Wired with expert communities, and Flipboard with creative users and communities. Foreign magazine contents diverge into (paper, electronic, app and web magazine) as they start the lively trades of their contents on the magazine platform. In the area of contents uses, readers employ smart media technology effectively such as cloud computing, artificial intelligence and module individualization, making it possible for the virtuous cycle to remain in the relationship within communities, expert communities and creative users.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.4
• /
• pp.580-586
• /
• 2021

The CAT methodology is a numerical analysis technique using CAE. Recently, a methodology of applying artificial intelligence techniques to a simulation has been studied. A previous study compared the deformation results according to the injection molding process using a machine learning technique. Although MLP has excellent prediction performance, it lacks an explanation of the decision process and is like a black box. In this study, data was generated using Autodesk Moldflow 2018, an injection molding analysis software. Several Machine Learning Algorithms models were developed using RapidMiner version 9.5, a machine learning platform software, and the root mean square error was compared. The decision-tree showed better prediction performance than other machine learning techniques with the RMSE values. The classification criterion can be increased according to the Maximal Depth that determines the size of the Decision-tree, but the complexity also increases. The simulation showed that by selecting an intermediate value that satisfies the constraint based on the changed position, there was 7.7% improvement compared to the previous simulation.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.4
• /
• pp.105-112
• /
• 2021

In the era of big data, interest in data is exploding. In particular, the development of the Internet and social media has led to the creation of new data, enabling the realization of the era of big data and artificial intelligence and opening a new chapter in convergence technology. Also, in the past, there are many demands for analysis of data that could not be handled by programs. In this paper, an analysis model was designed and verified for classification of unstructured data, which is often required in the era of big data. Data crawled DBPia's thesis summary, main words, and sub-keyword, and created a database using KoNLP's data dictionary, and tokenized words through morpheme analysis. In addition, nouns were extracted using KAIST's 9 part-of-speech classification system, TF-IDF values were generated, and an analysis dataset was created by combining training data and Y values. Finally, The adequacy of classification was measured by applying three analysis algorithms(random forest, SVM, decision tree) to the generated analysis dataset. The classification model technique proposed in this paper can be usefully used in various fields such as civil complaint classification analysis and text-related analysis in addition to thesis classification.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.12
• /
• pp.63-71
• /
• 2020

In this paper, we propose a fire detection technology using YOLOv3 and EfficientDet, the most reliable artificial intelligence detection algorithm recently, an alert service that simultaneously transmits four kinds of notifications: text, web, app and e-mail, and an AWS system that links fire detection and notification service. There are two types of our highly accurate fire detection algorithms; the fire detection model based on YOLOv3, which operates locally, used more than 2000 fire data and learned through data augmentation, and the EfficientDet, which operates in the cloud, has conducted transfer learning on the pretrained model. Four types of notification services were established using AWS service and FCM service; in the case of the web, app, and mail, notifications were received immediately after notification transmission, and in the case of the text messaging system through the base station, the delay time was fast enough within one second. We proved the accuracy of our fire detection technology through fire detection experiments using the fire video, and we also measured the time of fire detection and notification service to check detecting time and notification time. Our AI fire detection and notification service system in this paper is expected to be more accurate and faster than past fire detection systems, which will greatly help secure golden time in the event of fire accidents.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.3
• /
• pp.133-139
• /
• 2021

Currently, the process of constructing a high-definition road map has a high proportion of manual labor, so there are limitations in construction time and cost. Research to automate map production with high-definition road maps using artificial intelligence is being actively conducted, but since the construction of training data for the map construction is also done manually, there is a need to automatically build training data. Therefore, in this study, after converting to images using point clouds acquired by a mobile mapping system, the road marking areas were extracted through image reclassification and overlap analysis using thresholds. Then, a methodology was proposed to automatically construct training data for deep learning data for the high-definition road map through the classification of the polygon types in the extracted regions. As a result of training 2,764 lane data constructed through the proposed methodology on a deep learning-based PointNet model, the training accuracy was 99.977%, and as a result of predicting the lanes of three color types using the trained model, the accuracy was 99.566%. Therefore, it was found that the methodology proposed in this study can efficiently produce training data for high-definition road maps, and it is believed that the map production process of road markings can also be automated.

• Journal of the Korean Society of Radiology
• /
• v.16 no.5
• /
• pp.595-602
• /
• 2022

This study tried to suggest the most efficient learning rate for accurate and efficient automatic diagnosis of medical images for chest X-ray pneumonia images using deep learning. After setting the learning rates to 0.1, 0.01, 0.001, and 0.0001 in the Inception V3 deep learning model, respectively, deep learning modeling was performed three times. And the average accuracy and loss function value of verification modeling, and the metric of test modeling were set as performance evaluation indicators, and the performance was compared and evaluated with the average value of three times of the results obtained as a result of performing deep learning modeling. As a result of performance evaluation for deep learning verification modeling performance evaluation and test modeling metric, modeling with a learning rate of 0.001 showed the highest accuracy and excellent performance. For this reason, in this paper, it is recommended to apply a learning rate of 0.001 when classifying the presence or absence of pneumonia on chest X-ray images using a deep learning model. In addition, it was judged that when deep learning modeling through the application of the learning rate presented in this paper could play an auxiliary role in the classification of the presence or absence of pneumonia on chest X-ray images. In the future, if the study of classification for diagnosis and classification of pneumonia using deep learning continues, the contents of this thesis research can be used as basic data, and furthermore, it is expected that it will be helpful in selecting an efficient learning rate in classifying medical images using artificial intelligence.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_3
• /
• pp.913-924
• /
• 2022

In this paper, we presented a database construction of undersea images for the Habitats of Ecklonia cava and Sargassum and conducted an experiment for semantic segmentation using state-of-the-art (SOTA) models such as High Resolution Network-Object Contextual Representation (HRNet-OCR) and Shifted Windows-L (Swin-L). The result showed that our segmentation models were superior to the existing experiments in terms of the 29% increased mean intersection over union (mIOU). Swin-L model produced better performance for every class. In particular, the information of the Ecklonia cava class that had small data were also appropriately extracted by Swin-L model. Target objects and the backgrounds were well distinguished owing to the Transformer backbone better than the legacy models. A bigger database under construction will ensure more accuracy improvement and can be utilized as deep learning database for undersea images.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.4
• /
• pp.11-19
• /
• 2021

An abnormal object refers to a person, an object, or a mechanical device that performs abnormal and unusual behavior and needs observation or supervision. In order to detect this through artificial intelligence algorithm without continuous human intervention, a method of observing the specificity of temporal features using optical flow technique is widely used. In this study, an abnormal situation is identified by learning an algorithm that translates an input image frame to an optical flow image using a Generative Adversarial Network (GAN). In particular, we propose a technique that improves the pre-processing process to exclude unnecessary outliers and the post-processing process to increase the accuracy of identification in the test dataset after learning to improve the performance of the model's abnormal behavior identification. UCSD Pedestrian and UMN Unusual Crowd Activity were used as training datasets to detect abnormal behavior. For the proposed method, the frame-level AUC 0.9450 and EER 0.1317 were shown in the UCSD Ped2 dataset, which shows performance improvement compared to the models in the previous studies.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.7 no.2
• /
• pp.199-208
• /
• 2017

IoT, Cloud, Bigdata, Mobile, AI, and 3D print, which are called as the main axis of the 4th Industrial Revolution, can be predicted to be changed when the technology is applied to the military. Especially, when I think about the purpose of battle, I think that IoT, Cloud, Bigdata, Mobile, and AI will play many role. Therefore, in this paper, Smart Military is defined as the future military that incorporates these five technologies, and the architecture is established and the appropriate information security model is studied. For this purpose, we studied the existing literature related to IoT, Cloud, Bigdata, Mobile, and AI and found common elements and presented the architecture accordingly. The proposed architecture is divided into strategic information security and tactical information security in the Smart Military environment. In the case of vulnerability, the information security is divided into strategic information security and tactical information security. If a protection system is established, it is expected that the optimum information protection can be constructed within an effective budget range.