Vol. 46, No. 5,
May 2019
Digital Library
Fast Blockchain Consensus Algorithm with Forward Secure Signatures
Jeonghyuk Lee, Jihye Kim, Hyunok Oh
http://doi.org/10.5626/JOK.2019.46.5.397
Recently blockchain has emerged as an alternative to central data management. Existing blockchains, such as Bitcoin or Ethereum use a PoW(Proof of Work) method to reliably add a new block to the blockchain. Since PoW method performs a hash function calculation and has a high computational cost, fast transactions are impossible with PoW. Therefore, we propose a delegation based blockchain that can replace PoW method and use a forward secure signature to enhance the blockchain security. We implemented a signature scheme that could be used in delegation based blockchains, and analyzed the performance and security of the proposed blockchain.
Retrieval-Based Hair Model Augmentation for 3D Face Modeling
http://doi.org/10.5626/JOK.2019.46.5.405
Building a 3D head model from images is a good strategy that can be used in various fields such as facial analysis, face animation, and biometric recognition. In particular, the part of the head with hair has a variety of shapes and increased complexity compared to face. However, previous studies mainly focused on 3D face modeling for only facial region, leaving out hair modeling. In this paper, a method to retrieve the most similar 3D hair model from the database is proposed. Projected 3D hair models are compared with the segmented hair region which is extracted using FCN (fully connected network). The retrieved 3D hair model is augmented on 3D face model so that a complete head is modeled with face and hair. The proposed method includes network training and 3D hair database building. Experimental result shows that proper hair models are augmented successfully on the 3D face modeling result.
Comparative Analysis of Various Korean Morpheme Embedding Models using Massive Textual Resources
http://doi.org/10.5626/JOK.2019.46.5.413
Word embedding is a transformation technique that enables a computer to recognize natural language. It is used in various fields of natural language processing based on machine learning such as machine translation and named-entity recognition. Various word-embedding models are available; however, few studies have compared the performance of these models under similar conditions. In this paper, we compare and analyze the performance of Word2Vec Skip-Gram, CBOW, Glove, and FastText, which are actively used according to Korean morpheme spacing. Based on experimental results with large news corpus and Sejong corpus, FastText yielded the best performance among CBOW, Skip-gram, Glove, and FastText of Word2Vec.
Implementation of a Stable Point-of-View for Dual Gazing based on the Principle of Eye Movement
Ire Eom, Hanna Lee, Adithya B, Youngho Chai
http://doi.org/10.5626/JOK.2019.46.5.419
Using a dual gaze with eye movement in a first-person game provides a stable view while minimizing blurring of images. The First Person Shooter (FPS) game is based on a first-person viewpoint, since the camera responsible for the player`s gaze and the characters in the game are combined. Thus, when the character moves, the viewpoint moves together, which is the same principle underlying fixed vision with a still head. However, the eyes and the head are separate. Therefore, under multiple eye movements, the head follows the eye after gazing at the object initially, which allows a steady gaze at an object even under body movements.
In this paper, we propose a stable perspective based on the principle of vestibular reflex to the camera in the FPS game. The game environment is created using the Unity game engine, and the stable viewpoint of the dual gaze is demonstrated by comparing the viewpoints when the eye and head are fixed under specific scenarios and during the vestibular reflex.
Scene Generation from a Sentence by Learning Object Relation
Yongmin Shin, Su Jeong Choi, Seong-Bae Park, Seyoung Park
http://doi.org/10.5626/JOK.2019.46.5.431
In communication between humans and machines, location information is crucial. However, it is sometimes omitted. While humans can infer omitted information, machines cannot. Thus, certain problems can occur when generating scenes from sentences. In order to solve this problem, previous studies have found an explicit relation in the sentence, then inferred an implicit relation by using prior probability. However, such methods are not suitable for Korean, as it has morphologically productivity. In this paper, we suggest a scene-generation method for Korean. Frist, we find an explicit relation by using an RNN-based artificial neural network. Then, to infer implicit information, we use the prior probability of relations. Finally, we prepare a scene tree with the obtained information, then generate a scene using that tree. In order to evaluate the scene generation, we measure the accuracy of the model dealing with the relationship and assign a human score to the generated scene. As a result, the method is proven to be effective with excellent performance and evaluation.
Semi-automatic Expansion for a Chatting Corpus Based on a K-means Clustering Method And Similarity Measure
http://doi.org/10.5626/JOK.2019.46.5.440
In this paper, we proposed a semi-automatic expansion method to expand a chatting corpus using a large amount of utterance data from movie subtitles and drama scripts. To expand the chatting corpus, the proposed system used previously constructed chatting corpus and a similarity measure. If the similarity is calculated between a previously constructed chatting corpus and the input utterance was greater than a threshold value set in the experiment, the input utterance was selected as a new chatting utterance, that it is a correct chatting pair. We used morpheme-unit word embeddings and a Convolutional Neural Networks to efficiently calculate the similarity of the utterance embedding. In order to improve the speed of the semi-automatic expansion process, we proposed to reduce the amount of computation by clustering chat corpus by K-means clustering algorithm. Experimental results showed that the precision, recall, and F1 score of the proposed system were 61.28%, 53.19%, and 56.94%, respectively, which was 5.16%p, 6.09%, and 5.73%p higher than that of the baseline system. The term frequency and the speed of our system were also about a hundred times faster.
Implementation of Software Source Code Obfuscation Tool for Weapon System Anti-Tampering
Gyuho Lee, Jaegwan Yu, Insung Kim, Taekyu Kim
http://doi.org/10.5626/JOK.2019.46.5.448
The increasing functional complexity and diversity of the weapon systems has reinforced the significance of weapon system software. However, as the range of software functions expands, the insertion of core algorithms and critical data into the weapon system execution binaries, and reverse engineering has facilitated hacking and tampering of such information with malicious intent. In this paper, we propose an obfuscation tool that utilizes obfuscation techniques against source code for the development of weapon system software. In particular, control flow obfuscation techniques were applied to obfuscate core algorithms, and data obfuscation techniques were proposed to conceal important data. In addition, considering the actual performance of the weapon system software, the system was implemented in a user-friendly and flexible structure for selection based on level. The experimental findings confirmed the performance of the techniques used. These source code-based obfuscation techniques can be used to create anti-reverse engineering binary files and to develop anti-tampering platforms for weapon system software in the future.
Mechanized Proof for Type Preservation of GTLC
http://doi.org/10.5626/JOK.2019.46.5.457
Over the past decade, there has been a growing trend of supporting selective use of static typing in dynamic languages, coined "gradual typing", among the major real-world programming languages (e.g., Python, JavaScript), motivating studies on the theoretical basis for gradual typing. We developed mechanized proof for subject reduction, which is one of the two key parts of the type safety property of the gradually typed lambda calculus(GTLC), using the Abella proof assistant. We further elaborate on some parts of our mechanized proofs in order to highlight the pros and cons of using Abella for developing mechanized theories on gradual typing, compared to prior related work on GTLC using other proof assistants.
Elastic Multiple Parametric Exponential Linear Units for Convolutional Neural Networks
http://doi.org/10.5626/JOK.2019.46.5.469
Activation function plays a major role in determining the depth and non-linearity of neural networks. Since the introduction of Rectified Linear Units for deep neural networks, many variants have been proposed. For example, Exponential Linear Units (ELU) leads to faster learning as pushing the mean of the activations closer to zero, and Elastic Rectified Linear Units (EReLU) changes the slope randomly for better model generalization. In this paper, we propose Elastic Multiple Parametric Exponential Linear Units (EMPELU) as a generalized form of ELU and EReLU. EMPELU changes the slope for the positive part of the function argument randomly within a moderate range during training, and the negative part can be dealt with various types of activation functions by its parameter learning. EMPELU improved the accuracy and generalization performance of convolutional neural networks in the object classification task (CIFAR-10/100), more than well-known activation functions.
Reverse Collective Spatial Keyword Queries based on G-tree for Road Networks
http://doi.org/10.5626/JOK.2019.46.5.478
With the proliferation of mobile devices and social network services, spatial keyword queries have become a hot research topic. Previous works have focused on collective spatial keyword queries (CSKQs), which find a set of objects that covers the queried keywords and is close to the query location. In addition, analyzing the correlation between two objects has been studied extensively for real-world applications, such as location recommendations, personalized advertisements, and online social marketing services. CSKQ is suitable for supporting these services because it returns a correlated set of objects. However, the existing studies on CSKQ have focused only on the users’ perspective, despite the fact that such applications require the objects’ perspective. To address this problem, we propose a novel spatial keyword query (reverse collective spatial keyword query, RCSKQ) and a query processing technique based on the road network environment with a G-tree index structure.
Research for Speed Improvement Method of Lightweight Block Cipher CHAM using NEON SIMD
Sujin Lee, Junyoung Kang, Dowon Hong, Changho Seo
http://doi.org/10.5626/JOK.2019.46.5.485
As embedded devices and IoT devices are being developed, lightweight block ciphers have been proposed to achieve confidentiality on low-end devices. Recently, a lightweight block cipher algorithm, called CHAM, with 4-branch Feistel structure was proposed in Korea. It is consists of CHAM-64/128, CHAM-128/128, and CHAM-128/256 depending on the size of plaintext and secret key. CHAM, which is based on ‘stateless on the fly’ key schedule and ARX operations, is efficient on embedded devices, especially on low-end devices. In this paper, we analyze the lightweight block cipher CHAM and study an optimization method on a high-end IoT device. We implemented a serial code by independently generating round keys and leveraging 4-branch Feistel structure, and optimized CHAM using NEON-SIMD on ARM Cortex-A53.
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