Vol. 46, No. 6,
Jun. 2019
Digital Library
Design and Implementation of 10-Bit Wallace Tree Multiplication using the Modified 4-3 Compressor and the Reduction Method
http://doi.org/10.5626/JOK.2019.46.6.493
Multipliers play an important role in almost all high performance digital systems. The implementation of a conventional Wallace Tree multiplication is based on the CSAs (Carry Saved Adders). This paper proposes a new implementation method of high performance multiplier using CSAs, modified 4-3 compressors and the method of reduction, while using FPGA to implement a small AI (Artificial Intelligence) networks. Particularly, it is implemented through the Verilog-HDL and verified with the Xilinx ISE Design suite 14.7. The proposed design reduces the latency by 19.1% compared to the traditional Wallace Tree multiplication and by 6.2% in comparison to the method of reduction.
An Evaluation of NVMe-over-Fabrics on PCIe JBOF SSDs
Jake Choi, Heon Y. Yeom, Hyuck Han
http://doi.org/10.5626/JOK.2019.46.6.499
NVMe-over-Fabrics allows the management of target server storage devices directly from the host machine through a network fabric. Storage devices do not have to be connected directly to the PCIe bus allowing for storage disaggregation from the compute resources. As the total capacity of storage devices connected to target machines is increased by PCIe JBOFs, there could be more concurrent I/O operations from host machines to target NVMe storage devices in real-life workloads. This article shows that concurrent I/O operations from host machines to several NVMe storage devices over NVMe-over-Fabrics suffer a heavy performance loss compared to concurrent I/O on directly attached storage devices.
Design of Photovoltaic Power Generation Prediction Model with Recurrent Neural Network
Hanho Kim, Haesung Tak, Hwan-gue Cho
http://doi.org/10.5626/JOK.2019.46.6.506
The Smart Grid predicts the power generation amount of renewable energy and enables efficient power generation and consumption. Existing PV power generation prediction studies have rarely applied and compared recurrent neural network techniques that are superior to time series. Furthermore, in the reported studies, there is no consideration of the length of past data used for learning, leading to lowered prediction performance of the model. In this study, we used the embedded variable selection techniques to find the factors influencing PV power generation. Subsequently, experiments were carried out to insert various past data length into the recurrent neural networks (RNN, LSTM, GRU). We found the optimal prediction factors and designed a prediction model based on the outcomes of the experiments. The designed PV power generation prediction model shows better prediction performance compared to other factor settings. In addition, better performance based on the prediction rate is confirmed in the present study as compared with the existing researches.
Evaluation of Korean Reviews Automatically Generated using Long Short-Term Memory Unit
http://doi.org/10.5626/JOK.2019.46.6.515
Opinion Spam is the act of misleading the public to comply with the intent of a particular group. This is a serious problem at this point in time, when online services are widely deployed and the reputation of a specific target seller relative to the offline market can be easily affected. In this context, detecting and preventing opinion spam is an important research topic. This study investigates the current status of the domestic online market and poses possible damages caused by opinion spam through the automatic generation of Korean reviews using a recurrent neural network (RNN). In particular, we applied a learning model based on an LSTM (long short-term memory) unit for the Korean language in order to improve the long-term dependency problem of the typical RNN. Then, the Word2Vec is exploited to improve the relevance by replacing keywords with a set of candidates from the target market. We show the translation of the learning model based on a foreign language to Korean, then evaluate its appropriateness with regard to the development of countermeasures with which to effectively prevent automatically generated opinion spam in the near future.
An Educational Relay Programming System for Cooperative Programming Activities and Evaluations
http://doi.org/10.5626/JOK.2019.46.6.526
Recently, programming education has become important all around the world. Previously, programming education mostly involved individual ways of learning. However, the process of solving complex real-world problems through programming often requires grouping and cooperative programming. Therefore, a system is needed that enables cooperative programming and evaluation in programming education. Few educational systems exist for cooperative programming education. Meanwhile, a teacher must perform a process-oriented evaluation in order to maximize the educational effects of cooperative learning. But many teachers have felt burdened by this task. In this study, we designed and implemented a relay programming learning system to support cooperative programming learning. We developed factors to be considered in order to support the evaluation of an instructor and developed an automated evaluation algorithm. This allows learners to develop the ability to improve readability, reusability, and maintenance that are important in programming, and allows instructors to easily evaluate learners.
Solving for Redundant Repetition Problem of Generating Summarization using Decoding History
Jaehyun Ryu, Yunseok Noh, Su Jeong Choi, Seyoung Park, Seong-Bae Park
http://doi.org/10.5626/JOK.2019.46.6.535
Neural attentional sequence-to-sequence models have achieved great success in abstractive summarization. However, the model is limited by several challenges including repetitive generation of words, phrase and sentences in the decoding step. Many studies have attempted to address the problem by modifying the model structure. Although the consideration of actual history of word generation is crucial to reduce word repetition, these methods, however, do not consider the decoding history of generated sequence. In this paper, we propose a new loss function, called ‘Repeat Loss’ to avoid repetitions. The Repeat Loss directly prevents the model from repetitive generation of words by giving a loss penalty to the generation probability of words already generated in the decoding history. Since the propose Repeat Loss does not need a special network structure, the loss function is applicable to any existing sequence-to-sequence models. In experiments, we applied the Repeat Loss to a number of sequence-to-sequence model based summarization systems and trained them on both Korean and CNN/Daily Mail summarization datasets. The results demonstrate that the proposed method reduced repetitions and produced high-quality summarization.
Biomedical Named Entity Recognition using Multi-head Attention with Highway Network
Minsoo Cho, Jinuk Park, Jihwan Ha, Chanhee Park, Sanghyun Park
http://doi.org/10.5626/JOK.2019.46.6.544
Biomedical named entity recognition(BioNER) is the process of extracting biomedical entities such as diseases, genes, proteins, and chemicals from biomedical literature. BioNER is an indispensable technique for the extraction of meaningful data from biomedical domains. The proposed model employs deep learning based Bi-LSTM-CRF model which eliminates the need for hand-crafted feature engineering. Additionally, the model contains multi-head attention to capture the relevance between words, which is used when predicting the label of each input token. Also, in the input embedding layer, the model integrates character-level embedding with word-level embedding and applies the combined word embedding into the highway network to adaptively carry each embedding to the input of the Bi-LSTM model. Two English biomedical benchmark datasets were employed in the present research to evaluate the level of performance. The proposed model resulted in higher f1-score compared to other previously studied models. The results demonstrate the effectiveness of the proposed methods in biomedical named entity recognition study.
Grammatical Error Detection for L2 Learners Based on Attention Mechanism
Chanhee Park, Jinuk Park, Minsoo Cho, Sanghyun Park
http://doi.org/10.5626/JOK.2019.46.6.554
Grammar Error Detection refers to the work of discovering the presence and location of grammatical errors in a given sentence, and is considered to be useful for L2 learners to learn and evaluate the language. Systems for grammatical error correction have been actively studied, but there still exist limitations such as lack of training corpus and limited error type correction. Therefore, this paper proposes a model for generalized grammatical error detection through the sequence labeling problem which does not require the determination of error type. The proposed model dynamically decides character-level and word-level representation to deal with unexpected words in L2 learners" writing. Also, based on the proposed model the bias which can occur during the learning process with imbalanced data can be avoided through multi-task learning. Additionally, attention mechanism is applied to efficiently predict errors by concentrating on words for judging errors. To validate the proposed model, three test data were used and the effectiveness of the model was verified through the ablation experiment.
Jamo Unit Convolutional Neural Network Based Automatic Classification of Frequently Asked Questions with Spelling Errors
Youngjin Jang, Harksoo Kim, Dongho Kang, Sebin Kim, Hyunki Jang
http://doi.org/10.5626/JOK.2019.46.6.563
Web and mobile users obtain the desired information using the frequently asked questions (FAQ) listed on the homepage. The FAQ system displays a query response candidate that is most similar to the input based on an information retrieval model. However, the information retrieval model depends on the index, and therefore, it is vulnerable to spelling errors in the sentence. This paper proposes a model applying the FAQ system to the sentence classifier, which minimizes the spelling errors. Using the embedded layer with jamo-based convolutional neural network, the spelling errors of the user input were reduced. The performance of the classifier was improved using class embedding and feed-forward neural network. As a result of 457 and 769 FAQ classifications, the Micro F1 score showed 81.32% p and 61.11% p performance, respectively. We used the sigmoid function to quantify the reliability of the model prediction.
Improving the Upper Bound of the Dynamic Time Warping for Sparse and Long Time Sequences
Janghyuk Seo, Woohwan Jung, Kyuseok Shim
http://doi.org/10.5626/JOK.2019.46.6.570
Dynamic Time Warping (DTW), a distance measure widely used in time series analysis, is associated with the shortcoming of long execution time with long time sequences. To alleviate the problem, several algorithms which use a compression technique called run-length encoding and compute approximate DTW distances were recently proposed. However, the computation of the upper bounds by such algorithms consists of adding unnecessary distance values. In this paper, we propose an approximation algorithm which improves the state of the art for computing approximate DTW distances while keeping the time complexity unchanged. Experimental results with both synthetic and real-life data confirm the effectiveness of our proposed algorithm.
Traffic Steering System with Dual Connectivity for Video Streaming Services
Gi Seok Park, Hyunmin Noh, Jae Jun Ha, Hyung Jun Kim, Sang Heon Shin, Dong Hyun Kim, Jong Hwan Ko, Jeung Won Choi, Hwangjun Song
http://doi.org/10.5626/JOK.2019.46.6.577
In this paper, we propose a traffic steering system with dual connectivity to provide stable video streaming services for users by steering portion of the macrocell traffic into small cells. The proposed system achieves a good balance between fairness and social welfare in terms of video quality by allocating the radio resource of the macro base station. The user data flow is divided into two channels toward the macro base station and the small cell AP, and the users receive their data from both. In the proposed system, the fountain code is adopted to overcome practical issues in the dual connectivity. Moreover, the SDN is employed not only to rapidly react to time-varying network condition, but also to control network resources efficiently. The proposed system is implemented using NS-3. The simulation results show that the proposed system can achieve much better performance compared with existing traffic steering algorithms.
A Robust Three-Factor User Authentication Scheme based on Elliptic Curve Cryptography and Fuzzy Extractor
Trung Thanh Ngo, Tae-Young Choe
http://doi.org/10.5626/JOK.2019.46.6.587
A three-factor user authentication is appropriate to ensure a high degree of authentication. Fan and Lin proposed a typical three-factor authentication scheme, which requires token, password, and fingerprint. The scheme does not allow authentication in the absence of any of the three factors. Unfortunately, Fan and Lin"s scheme is associated with security risks such as vulnerability to insider attacks, stolen-verifier attacks, and message modification attacks. Yeh et al. proposed a three-factor user authentication, which overcomes such pitfalls and improves security and performance using elliptic curve cryptography. We found that Yeh et al."s scheme is still vulnerable to user impersonation attacks and server masquerading attacks. We propose a robust three-factor authentication scheme entailing server smart cards, elliptic curve cryptography, and a fuzzy extractor that address the foregoing flaws and result in enhanced security. The proposed scheme is resistant to various attacks and improves system performance. BAN logic is used to prove that the scheme establishes a secure channel.
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