Digital Library[ Search Result ]
Sentence Similarity Prediction based on Siamese CNN-Bidirectional LSTM with Self-attention
Mintae Kim, Yeongtaek Oh, Wooju Kim
http://doi.org/10.5626/JOK.2019.46.3.241
A deep learning model for semantic similarity between sentences was presented. In general, most of the models for measuring similarity word use level or morpheme level embedding. However, the attempt to apply either word use or morpheme level embedding results in higher complexity of the model due to the large size of the dictionary. To solve this problem, a Siamese CNN-Bidirectional LSTM model that utilizes phonemes instead of words or morphemes and combines long short term memory (LSTM) with 1D convolution neural networks with various window lengths that bind phonemes is proposed. For evaluation, we compared our model with Manhattan LSTM (MaLSTM) which shows good performance in measuring similarity between similar questions in the Naver Q&A dataset (similar to Kaggle Quora Question Pair).
Evaluation of Interest Point Detectors for Data Authentication in Wireless Multimedia Sensor Network (WMSN)
http://doi.org/10.5626/JOK.2019.46.2.184
In Wireless Multimedia Sensor Networks (WMSNs), authentication of multimedia data is very important because the data can be used in making crucial decisions. This study evaluates interest point detectors in terms of resilience to channel error occurred in WMSNs, robustness to JPEG compression, and sensitivity to image tampering. SIFT, SURF, ORB, AKAZE, SADDLE and HOG were evaluated with USC-SIPI image database by computing recall and precision between the original images and modified images by channel errors and JPEG compression and tampering. In addition, median filter and Gaussian filter were applied to reduce channel error and quantization errors from JPEG compression respectively and produced significant performance. AKAZE showed best performance for all conditions of experiments. The evaluation of interest point detectors showed the possibility of their application to authentication in WMSNs.
GAN considering ERF for High-resolution Map Generation
http://doi.org/10.5626/JOK.2019.46.2.122
The paper proposes a network structure for a generative adversarial network (GAN) suitable for high resolution image transformation. For analysis of the resolution classification relation necessary for high resolution image conversion, the effective size of the receptive fields of each encoder is calculated and new connection imbalance fields defined. We can reduce the total number of layers by connecting the encoder and decoder to the patch size, we reduce the total number of layers and the appropriate effective receptive fields and parameter usability confirmed through experiments. To solve the problem of simultaneously providing resolution and classification in high resolution image conversion, a network structure capable of converting high resolution satellite images is suggested experimentally. Additionally, the validity of the network structure that simultaneously improves the resolution and classification is confirmed by comparing and analyzing the receptive fields of the proposed network and the existing network’s receptive fields. The proposed network is then quantitatively verified by comparing the proposed network with the existing network by use of objective numerical value through SSIM, an image similarity analysis method.
Korean Machine Reading Comprehension with S²-Net
Cheoneum Park, Changki Lee, Sulyn Hong, Yigyu Hwang, Taejoon Yoo, Hyunki Kim
http://doi.org/10.5626/JOK.2018.45.12.1260
Machine reading comprehension is the task of understanding a given context and identifying the right answer in context. Simple recurrent unit (SRU) solves the vanishing gradient problem in recurrent neural network (RNN) by using neural gate such as gated recurrent unit (GRU), and removes previous hidden state from gate input to improve speed. Self-matching network is used in r-net, and this has a similar effect as coreference resolution can show similar semantic context information by calculating attention weight for its RNN sequence. In this paper, we propose a S²-Net model that add self-matching layer to an encoder using stacked SRUs and constructs a Korean machine reading comprehension dataset. Experimental results reveal the proposed S²-Net model has EM 70.81% and F1 82.48% performance in Korean machine reading comprehension.
Efficient Method of Collecting Network Traces for Generating Network Topology
http://doi.org/10.5626/JOK.2018.45.12.1319
Network topology information is critical in cyber security for designing security architecture and threat analysis as well as for network management and diagnosis. Numerous approaches have been proposed for obtaining information about network topology. In particular, graph analytical methods for inferring network topology are intensively researched. These methods collect path traces via traceroute and analyze them using graph theoretical methods for inferring network topology. However, there exist few research reports on choosing destinations and deployment locations of trace collectors which have the potential of significantly affecting network overhead and discovery time. This paper proposes a novel method of choosing destinations and determining trace collectors for the efficient collection of network traces. In the present work, we have also implemented a prototype of the proposed methods and experimentally validated their performance.
Detection of Malicious Users with High Influence through Foul Language Network Analysis in MOBA Games
http://doi.org/10.5626/JOK.2018.45.12.1312
In relation to the online game industry, verbal violence in the game has become a serious social problem. However, it is difficult to solve fundamental problems by simply filtering or using reporting systems. This study proposed a method to analyze the propagation tendency of the foul language and to detect malicious users in social network perspective. This method was applied to the analysis of the chat log of Defense of the Ancients 2(DotA 2), a popular MOBA(Multiplayer Online Battle Arena) genre game around the world. In the case of MOBA games, there are usually limited users belonging to one queue, which is a good platform for analyzing foul language networks as compared to other games. Verbally abusive malicious users tend to have high centrality when they form a network. Using these features, we analyzed the propagation tendency of the foul language on the network and detected users with high centrality. We also analyzed the effect on the whole network when the user was restricted. With the proposed method, we were able to detect malicious users who used the foul language. For future works, we will classify the spreading types in the foul language network and analyze users for each type.
Collecting Network Field Information using Machine Learning
Kyu Seok Han, Taekyu Kim, Shinwoo Shim, Sung Goo Jun, Jiwon Yoon
http://doi.org/10.5626/JOK.2018.45.10.1096
Recently, various systems based on Internet of Things (IOT) and Information and Communications Technologies(ICT) have been developed. Today, assorted devices are connected to a network, and various operating systems according to devices having different resources and functions have appeared. With the increased need for in hacking security, researches on the vulnerability analysis of the operating system installed on each device and the actual attack technique have been carried out. Accordingly, the type and detailed version of the operating system of the device, Function (API) is emerging as important information in security. Since the control of this information gathering in the cyber warfare is the first stage of the cyber threat, many studies have been conducted on mehods for controlling the network traffic while scanning. In order to bypass this control of the network, information collectors prepare countermeasures to secretly collect port information. In this paper, we deal with a scanning method that can acquire information about opponents through network basic commands which are not important in the network control system.
Backbone Network for Object Detection with Multiple Dilated Convolutions and Feature Summation
Vani Natalia Kuntjono, Seunghyun Ko, Yang Fang, Geunsik Jo
http://doi.org/10.5626/JOK.2018.45.8.786
The advancement of CNN leads to the trend of using very deep convolutional neural network which contains more than 100 layers not only for object detection, but also for image segmentation and object classification. However, deep CNN requires lots of resources, and so is not suitable for people who have limited resources or real time requirements. In this paper, we propose a new backbone network for object detection with multiple dilated convolutions and feature summation. Feature summation enables easier flow of gradients and minimizes loss of spatial information that is caused by convolving. By using multiple dilated convolution, we can widen the receptive field of individual neurons without adding more parameters. Furthermore, by using a shallow neural network as a backbone network, our network can be trained and used in an environment with limited resources and without pre-training it in ImageNet dataset. Experiments demonstrate we achieved 71% and 38.2% of accuracy on Pascal VOC and MS COCO dataset, respectively.
RANSAC-based Time Synchronization over Wireless Networks
Junyoup Hwang, Daehyeon Wi, Jungjin Lee
http://doi.org/10.5626/JOK.2018.45.7.626
This paper describes a method to improve the quality of time synchronization for synchronous video streaming over a wireless network using a Simple Network Time Protocol (SNTP) with the RANdom SAmple Consensus (RANSAC) Algorithm. Time synchronization is mostly accomplished by using the NTP server. Speed and accuracy are important in synchronizing the video play module; however, NTP is limited because too much time is spent to operate the module. It also requires a complex structure and algorithms for customization. And SNTP, which is easy to implement, is also inappropriate for the wireless network because there is no filtering process for network failure. Our method refined those timestamps of SNTP using RANSAC, so that it achieves a stable synchronization, with 5 milliseconds of accuracy, which takes less than 5 seconds. This method enables high- quality time synchronization with a simple implementation for any kind of system, and will be useful for IoT devices.
A Competitive Multipath Routing Protocol for delay-sensitive applications in Wireless Sensor Networks
http://doi.org/10.5626/JOK.2018.45.6.589
The key performance of various data delivery applications studied in wireless sensor networks is the timeliness and reliability of transmission. Both performances may be required simultaneously depending on the type of application and information. However, because of the limited nature of resources in wireless sensor networks, these requirements are not easy to meet. One way to overcome this problem is by multipath routing method. Traditional multi-path routing protocols exploit a method of generating additional independent paths or branching an existing path to satisfy the required performance. These methods can waste too many network resources in an irregular network environment. In order to solve this problem, this paper proposes an energy-efficient multipath routing method that can satisfy the requirement of emergency application by using the competition and cooperation between paths in irregular wireless sensor networks. Finally, this paper compares and analyzes the routing performance of the proposed method by means of simulation.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr