Earticle

딥러닝 모델을 활용한 인공신경망 기계번역 (Neural machine translation)이 주류 분야로 떠오르면서 최고의 성능을 위해 모델과 데이터 언어 쌍에 대한 많은 투자와 연구가 활발하게 진행되고 있다. 그러나, 최근 대부 분의 인공신경망 기계번역 연구들은 번역 문장의 품질을 극대화하는 자연어 생성을 위한 디코딩 전략 (Decoding strategy)에 대해서는 미래 연구 과제로 남겨둔 채 다양한 실험과 구체적인 분석이 부족한 상황이다. 기계번역에서 디코딩 전략은 번역 문장을 생성하는 과정에서 탐색 경로를 최적화 하고, 모델 변경 및 데이터 확장 없이도 성능 개선이 가능하다. 본 논문은 시퀀스 투 시퀀스 (Sequence to Sequence) 모델을 활용한 신경망 기반의 기계번역 에서 고전적인 그리디 디코딩 (Greedy decoding)부터 최신의 방법론인 Dynamic Beam Allocation (DBA)까지 비교 분석하여 디코딩 전략의 효과와 그 의의를 밝힌다.

Neural machine translation using deep neural network has emerged as a mainstream research, and an abundance of investment and studies on model structure and parallel language pair have been actively undertaken for the best performance. However, most recent neural machine translation studies pass along decoding strategy to future work, and have insufficient a variety of experiments and specific analysis on it for generating language to maximize quality in the decoding process. In machine translation, decoding strategies optimize navigation paths in the process of generating translation sentences and performance improvement is possible without model modifications or data expansion. This paper compares and analyzes the significant effects of the decoding strategy from classical greedy decoding to the latest Dynamic Beam Allocation (DBA) in neural machine translation using a sequence to sequence model.

PDF417은 2차원 바코드로서 사용범위가 넓고 방대한량을 압축하여 저장하는 능력을 가지고 있다. 이러한 특성을 가진 PDF417은 현재 중요한 문서의 위조나 변조를 막기 위한 다양한 용도로 쓰이고 있다. 한편 기존의 PDF417 바코드의 디코딩 기술은 단순히 기본 표준규격서인 AIM에서 제시한 디코딩하므로 비효율적이라 할 수 있다. 따라서 본 논문에서는 PDF417 바코드를 보다 빠르고 효율적인 디코딩 알고리즘을 제시하고자 한다. 결과적으로 제안하는 디코딩 알고리즘은 기존 방법과 비교하여 불필요한 디코딩 과정을 제거함으로서 빠르고 효율적인 디코딩 알고리즘이 될 것이다.

Two-dimensional PDF417 bar code has a wide range of use and has a storage capacity to compress a large amount of data. With these characteristics, PDF417 has been used in various ways to prevent the forgery and alteration of important information in documents. On the other hand, previous decoding methods in PDF417 barcode are slow and inefficient because they simply employ the standard specifications of AIM (Association for Automatic Identification and Mobility). Therefore, this paper propose an efficient and fast algorithm of decoding PDF417 bar code. As a result, the proposed decoding algorithm will be more faster and efficient than previous methods.

2차원 QR Code는 1차원 바코드의 용량, 크기, 방향에 대한 한계를 개선하였으며, 방대한량의 데이터를 압축하여 저장할 수 있는 능력을 갖추고 많은 응용분야에서 사용되고 있다. 최근에는 QR Code를 인식할 수 있는 스마트 폰의 도입과 사용의 증대로 QR Code의 도입이 급속도로 확산되었다. 하지만 다양한 정보를 저장할 수 있는 QR Code의 디코딩 기술에 대한 구체적인 문서화가 없는 상태이다. 따라서 본 논문에서는 QR code의 인코딩에 따른 디코딩 과정에 대해 구체적인 처리 절차와 알고리즘을 제시하고 예를 통해 디코딩 과정을 상세히 이해시키는 데 연구목적이 있다.

Two-dimensional QR Code has improved greatly the limits on the capacity, size, direction of the one-dimensional bar code. And it has the capacity of compressing and storing the massive amount of data and is widely used in many applications. Recently, the two-dimensional QR Code has been spread rapidly because of introducing the smart phones and increasing the amount of using them. However, there is little documentations about decoding the QR Code in which can store the large amount of information. In this paper, therefore, we present specific processing procedures and algorithms on decoding the two-dimensional QR Code and then make us understand their decoding process by explaining some examples.

기존의 바코드는 정보 배열이 나란히 나열된 선 모양을 가지며 이를 1차원 바코드라 부른다. 이에 반해 2차원 바코드는 점자방식 또는 모자이크방식 코드로 작은 정사각형 도는 직사각형 안에 정보를 표현한다. 2차원 바코드는 기존의 1차원 바코드보다 작은 공간에 많은 데이터를 표현 가능함으로써 보다 효율적인 바코드의 구현이 가능하다. 현재 ISO 국제 표준화된 2차원 바코드는 총 4가지로 분류되는데 QR Code, Data Matrix, PDF417, MaxiCode가 있다. 본 논문에서는 ISO 국제 표준화된 바코드 중 하나인 Data Matrix의 Base 256 모드에 대한 기본 개념, 구성 방법, 인코딩 및 디코딩 방법을 상세히 제안한다. Data Matrix 심벌에 저장된 데이터를 보다 효율 적으로 구성하기 위해 숫자, Alphanumeric 문자, 이진법에 따라 다른 인코딩, 디코딩 방법을 사용하게 되는데 본 논문에서는 이를 고려한 디코딩 방법에 초점을 맞춰 기술할 것이다.

Conventional bar code has the appearance of line bars and spaces, called as one-dimensional bar code. In contrast, the information in two-dimensional bar code is represented by either a small, rectangular or square with the types of mosaic and Braille. The two-dimensional bar code is much more efficient than one-dimensional bar code because it can allow to store and express large amounts of data in a small space and so far there is also a little information about decoding the Data Matrix in base 256 mode. According to the ISO international standards, there are four kinds of bar code: QR code, Data Matrix, PDF417, and Maxi code. In this paper, among them, we focus on describing the basic concepts of Data Matrix in base 256 mode, how to encode and decode them, and how to organize them in detail. In addition, Data Matrix can be organized efficiently depending on the modes of numeric, alphanumeric characters, and binary system and expecially, we focus on describing how to decode the Data Matrix code by four modes.

In order to identify the CAN message sent from the ECU, the CAN waveform and operating current were measured during the operation of the CAN driver.CAN ID was identified in the same phase as the measured voltage waveform and operating current, and reverse engineering was performed .And, by measuring CAN bus wire EMI noise, it was tested whether CAN ID was identified.CAN ID verified ECU data using CAN analyzer.

Hallyu has exerted a significant influence on the cultural lives of young Vietnamese audiences, with K-dramas serving as a meaningful medium for evoking memories and emotions. Despite the widespread popularity of Reply 1988 (2015) in Vietnam, scholarly research on how the series activates cultural memory and cultural proximity in audience reception remains limited. This study seeks to address this gap by examining the resonance of Reply 1988 through the perspectives of cultural memory and East Asian cultural proximity. Drawing on the theoretical frameworks of Assmann and Hall?Jauss, the research investigates how young Vietnamese viewers interpret the series and relate it to family memories, generational experiences, and contemporary social contexts. The dataset includes an online survey with 76 valid respondents aged 18?35, supplemented by follow-up interviews and an analysis of social media comments. The findings indicate that viewers do not approach Reply 1988 as a linear nostalgic text; instead, they actively reconstruct meaning through personal memories and cultural points of connection between Vietnam and Korea. The series functions as a “shared memory space” in which Vietnamese viewers reflect on their own lives through the narratives of 1980s Korea.

To elucidate the role and mechanism of microbes, we combined culture-dependent and culture- independent approaches to investigate differences in gut bacterial composition between sows and weaned pigs. Under anaerobic conditions, several nonselective and selective media were used for isolation from fecal samples. All isolated bacteria were identified and classified through 16S rRNA sequencing, and the microbiota composition of the fecal samples was analyzed by metagenomics using next generation sequencing (NGS) technology. A total of 278 and 149 colonies were acquired from the sow and weaned pig fecal samples, respectively. Culturomics analysis revealed that diverse bacterial genus and species belonged to Firmicutes, Actinobacteria, Proteobacteria, and Bacteroidetes were isolated from sow and weaned pigs. When comparing culture-dependent and culture-independent analyses, 191 bacterial species and 2 archaeal bacterial species were detected through culture-independent analysis, and a total of 23 bacteria were isolated through a culture-dependent approach, of which 65% were not detected by metagenomics. In conclusion, culturomics and metagenomics should be properly combined to fully understand the intestinal microbiota, and livestock-derived microbial resources should be informed by culturomic approaches to understand and utilize the mechanism of host-microbe interactions.

This letter proposes a powerful method to reconstruct the spectral image frames for increasing data transmission capability in Display Field Communication (DFC) systems. The proposed method reconstructs the spectral image iteratively using the good data symbol estimates fed back by the camera decoder. To begin with, initial image is estimated by means of known pilot symbols. Then, hard symbol decisions obtained by iterative decoding are used to improve the quality of the reconstructed image. The proposed technique selects good data pixels using hard information from the decoder and performs re-estimation of the image pixels. From numerical simulations, we show that the proposed method significantly boosts the data transmission capacity of the DFC system over the conventional DFC that uses reference frames.

In this study, we explore the potential of leveraging machine learning techniques, specifically auto-encoders (AE), for the decoding of linear block codes. Our findings suggest that this approach can outperform the conventional ordered statistics decoding (OSD) method, especially in a Rayleigh fading channel environment. We have rigorously trained the AE under both additive white Gaussian noise and Rayleigh fading channel conditions to ensure robustness in its performance. The output of the AE is combined with the received vector in a suitable manner to perform OSD. Through our experiments, we demonstrate that this proposed decoding approach yields better results than the conventional OSD method in Rayleigh fading channel when we used (23,12) Golay code.

The paper proposes reliable channel for medical uses via Physical Layer. The purpose is to serve medical data transmission from WBAN through the UMTS. This study is focusing on applying coding theory to carry high reliability and robustness, in order to satisfy various QoS requirements of medical data. The aim is to build Medical Network Channel that transmits WBAN medical data through the UMTS. The design of extra channel coding is considered as concatenated techniques. The method applies adaptive outer code with inner UMTS code. The accomplishment of QoS is investigated with numerical results proven under AWGN and Rayleigh fading.

This study examined public perceptions of generative AI through the analysis of user comments on YouTube news videos regrading generative AI from three major South Korean broadcasting networks. Using structural topic modeling, 56,708 comments from 105 news videos were analyzed. Among nine distinct topics that emerged, the top three prevalent topics centered on labor market change, AI control concerns, and service automation benefits. Temporal analysis revealed evolving discourse patterns. Employment-related concerns peaked after the release of ChatGPT's but subsequently declined; three topics gained increasing prominence including human-centered AI development, control concerns, and educational applications. These findings offer theoretical and practical implications for individuals, organizations, and institutions adopting generative AI.

The min-sum (MS) decoding for low-density parity-check codes, though less complex than the sum?product algorithm, suffers from worse error-correcting performance. For enhancement, neural MS decoders leveraging deep learning have recently been introduced, but how to train them has not been sufficiently discussed. In this paper, we propose a novel dataset construction method and also propose systematic learning strategies by finding a good combination of dataset composition, loss functions, weight sharing, weight assignment, and weight update method. Simulations demonstrate that the proposed method achieves better error-correcting performance than other works, especially in the error floor region, within a limited number of iterations.

In this paper, a bi-directional sliding window decoder is proposed for spatially coupled low-density parity-check (SC-LDPC) codes, which improves the decoding complexity and performance compared to the conventional sliding window decoding (SWD) by sharing messages at the overlapped part of forward and backward decoding windows. Moreover, by using proper scaling factors that determine the weight of each message at the overlapped part of two sliding windows, good local decoding effects can be efficiently spread out to both ends of SC-LDPC code during decoding process. Such effective message updates of the proposed bi-directional overlapped sliding window decoding (BO-SWD) improve error floor performance compared to the conventional SWD. The validity of BO-SWD is verified by simulation with various SC-LDPC ensembles.

Artificial Intelligence (AI) has been used extensively in all aspects of everyday life among people in recent times. Many techniques utilizing machine learning (ML) and deep learning (DL) models are being presented in this rapidly growing field of study. Most such models are generally regarded as “Black-Box” models since they are intrinsically complex and lack interpretable explanations for their decisions and conclusions. The lack of transparency increases the issue in the field of cybersecurity as implementing critical decisions in a system that cannot provide explanations for itself offers some evident risks. The lack of interpretability and transparency in existing AI techniques would make users distrust the models used to defend against cyberattacks, particularly given the increasingly complex and diverse nature of cyberattacks. Thus, Explainable Artificial Intelligence (XAI) must be utilized to construct cyber security models that are more understandable while keeping high accuracy and that enable users to understand, be reliable, and manage the future of cyber defence systems. This study provides a comprehensive survey of existing literature on using XAI to mitigate these challenges of cybersecurity black-box models. It emphasizes the significance of explainability in boosting faith and transparency in AI-driven systems and presents a thorough taxonomy of XAI techniques and technologies for cybersecurity applications. The study describes the evaluation criteria that are used to evaluate the effectiveness of XAI models, addresses different kinds of attacks like malware, phishing, and network intrusions, and shows how XAI techniques may mitigate these risks by providing a comprehensible understanding of model decisions. Along with the real-world case studies, it also explores the industrial applications of XAI in cybersecurity and examines the challenges in implementing XAI technology. The survey concludes with a review of the limitations of the existing XAI techniques and makes recommendations for future research, such as the requirement for more reliable XAI frameworks that can function in real-time and across a variety of cyber threat situations.

In this paper, we propose a novel successive cancellation flip (SCF) decoding to reduce the computational complexity compared to the conventional SCF decoding by using distributed CRC bits. The proposed decoding reduces the number of estimations for information bits by early termination of decodings for failed frames of the first SC decoding, while trying to minimize the additional sorting operations. Simulation results show that the proposed SCF decoding reduces the computational complexity of repeated SC decoding at least 27% compared to the conventional SCF decoding.

The application of social media for political campaigns is not a recent phenomenon among political marketing scholars. The social media campaigns of Barack Obama, Donald Trump, and Narendra Modi are among the well-studied cases in the political marketing literature. However, the motivations behind social media campaigns for political candidates in developing countries, such as Nepal, have largely been overlooked in the literature. This paper explores the social media engagement behaviors of three electoral candidates, drawing on social media engagement theory. It analyzes 582 Facebook posts from these candidates using a multi-method approach. The findings from content analysis, sentiment analysis, and thematic analysis reveal distinct messaging patterns, varied voter sentiments, and the motivations of political candidates. The motivations for these candidates’ engagement on social media during elections include fostering voter engagement, expressing gratitude for support, and intrinsic motivation. This research contributes unique insights into the social media engagement behaviors of electoral candidates, providing practical guidance for political campaign managers. Such insights can help create impactful content that resonates with voters, ultimately enhancing political communication on social media platforms.

The study aims to explore the nature, drivers of BAE. In particular, how and why consumers engage with branded mobile apps. Since consumer engagement is a multi-dimensional and complex phenomenon, the authors took an interpretivist paradigm and used a grounded theory approach by interviewing 24 participants. The findings reveal that branded app engagement (BAE) is triggered by mainly three aspects, i.e., media characteristics (Privacy and security, customization, and convenience), brand-related aspects (brand image and product/ service quality), and benefits sought (inherent and derived). More importantly, we uncover the underlying mechanism for each factor. Furthermore, the study explains the BAE process in its entirety from an app-centric perspective rather the brand powering the app. Thus, the study offers a fresh perspective and novel insights into the BAE process.

The earthquake of 2015 cause immense damage to the cultural heritage of Kathmandu Valley. It also regenerated the debates on authenticity. Due to the material use, seismic vulnerability and organic linkages of people with the heritage, the values defining authenticity were quite different. Rani Pokhari was chosen as a study site among several sites that were destroyed in the earthquake. This site is an interesting case from several aspects. First, it was the most controversial site under reconstruction and also the first site where the locals, experts and youth protested for the conservation in the traditional style. Second, it was a white washed dome shaped temple in the centre of the pond before the earthquake and was reconstructed in sikhara style. Also, from its construction in 1670 to 2015 it had gone through several restorations, renovations and even reconstruction. The study took place during the reconstruction by the National Reconstruction Authority so the on-site observation along with the controversies and issues raised could be included and analysed. Looking through the lenses of authenticity in heritage reconstruction, this paper explores different values in the context of Nepal and seeks to understand the traditional practices and the recent practice in heritage conservation post 2015.