Vol. 48, No. 12,
Dec. 2021
Digital Library
One-to-One Disjoint Cover of Non-bipartite k-ary n-cube with Faulty Edges
Hee-Chul Kim, Sang-Young Cho, Chan-Su Shin
http://doi.org/10.5626/JOK.2021.48.12.1251
Given a graph G, a source s, and a sink t, one-to-one r -disjoint path cover joining s and t is a set of r internally vertex disjoint paths joining s and t which cover all vertices in G. In this paper, we consider one-to-one disjoint path cover problem in the k-ary n-cube with faulty edges which is one of the well-known interconnection networks. It is shown that in non-bipartite k-ary n-cube Qnk with n ≥ 3, k ≥ 3 , and f faulty edges where f ≤ 2n-3, there is one-to-one r-disjoint path cover joining any two vertices for any r with f+r≤2n. The upper bound, 2n, on f+r is the best possible.
Analyses of Linux I/O Interfaces for High-Performance I/O Operations in Key-Value Stores
http://doi.org/10.5626/JOK.2021.48.12.1274
Recently, to utilize high-performance storage devices, various I/O interfaces have been studied. In particular, asynchronous I/O interfaces are now getting attention. However, those asynchronous I/O interfaces have the disadvantage of relatively higher latency and more CPU consumption than synchronous I/O interfaces. In this paper, we first analyze the characteristics of each I/O interface and then apply it to the key-value store to measure the performance at the application layer. The experiments show that the asynchronous I/O interface consumes 2.73× more CPU resources than the synchronous I/O interface, but delivers up to 2.42× higher IOPS depending upon the batching I/O size. In addition, experiments on key-value stores show up to 6.60× higher throughput and 61.09% lower latency when asynchronous I/O interfaces are used in the read-intensive workload.
GPT-2 for Knowledge Graph Completion
http://doi.org/10.5626/JOK.2021.48.12.1281
Knowledge graphs become an important resource in many artificial intelligence (AI) tasks. Many studies are being conducted to complete the incomplete knowledge graph. Among them, interest in research that knowledge completion by link prediction and relation prediction is increasing. The most talked-about language models in AI natural language processing include BERT and GPT-2, among which KG-BERT wants to solve knowledge completion problems with BERT. In this paper, we wanted to solve the problem of knowledge completion by utilizing GPT-2, which is the biggest recent issue in the language model of AI. Triple information-based knowledge completion and path-triple-based knowledge completion were proposed and explained as methods to solve the knowledge completion problem using the GPT-2 language model. The model proposed in this paper was defined as KG-GPT2, and experiments were conducted by comparing the link prediction and relationship prediction results of TransE, TransR, KG-BERT, and KG-GPT2 to evaluate knowledge completion performance. For link prediction, WN18RR, FB15k-237, and UMLS datasets were used, and for relation prediction, FB15K was used. As a result of the experiment, in the case of link prediction in the path- triple-based knowledge completion of KG-GPT2, the best performance was recorded for all experimental datasets except UMLS. In the path-triple-based knowledge completion of KG-GPT2, the model"s relationship prediction work also recorded the best performance for the FB15K dataset.
Layer-wise Relevance Propagation (LRP) Based Technical and Macroeconomic Indicator Impact Analysis for an Explainable Deep Learning Model to Predict an Increase and Decrease in KOSPI
http://doi.org/10.5626/JOK.2021.48.12.1289
Most of the research on stock prediction using artificial intelligence has focused on improving the accuracy. However, reliability, transparency, and equity of decision-making should be secured in the field of finance. This study proposes a layer-wise relevance propagation (LRP) approach to create an explainable stock prediction deep learning model, which is trained using macroeconomic and technical indicators as the input features. Also, the definition of the problem is simplified by prediction of an increase or decrease in the KOSPI closing price from the previous day instead of prediction of the KOSPI value itself. To show how the proposed method works, experiments are conducted. The results show that the model trained with data by the selected features via LRP is more accurate than the vanilla model. Moreover, we show that LRP results are meaningful by analyzing the tendency of the positive effect of each feature for the prediction results.
Data Augmentation Methods for Improving the Performance of Machine Reading Comprehension
Sunkyung Lee, Eunseong Choi, Seonho Jeong, Jongwuk Lee
http://doi.org/10.5626/JOK.2021.48.12.1298
Machine reading comprehension is a method of understanding the meaning and performing inference over a given text by computers, and it is one of the most essential techniques for understanding natural language. The question answering task yields a way to test the reasoning ability of intelligent systems. Nowadays, machine reading comprehension techniques performance has significantly improved following the recent progress of deep neural networks. Nevertheless, there may be challenges in improving performance when data is sparse. To address this issue, we leverage word-level and sentence-level data augmentation techniques through text editing, while minimizing changes to the existing models and cost. In this work, we propose data augmentation methods for a pre-trained language model, which is most widely used in English question answering tasks, to confirm the improved performance over the existing models.
Person Re-Identification Using an Attention Pyramid for Local Multiscale Feature Embedding Extracted from a Person’s Image
http://doi.org/10.5626/JOK.2021.48.12.1305
In this paper, a person re-identification scheme using the dual pyramid adapting attention mechanisms to extract more elaborate local feature embedding by excluding the noises caused by the unnecessary backgrounds in person’s image is proposed. With the dual pyramid of local and scale ones, the spatial attention is used to suppress the noise effects caused by unnecessary backgrounds, and the channel attention is used to emphasize the relatively important multiscale features when the local feature embedding is constructed. In the experiments, the proposed scheme was compared with other cases in which the attention module is not used for each pyramid to confirm the optimal configuration and compared based on the rank-1 accuracy with the state-of-the-art studies for the person re-identification. According to the experimental results, the proposed method showed a maximum rank-1 accuracy of 99.4%, which is higher by at least about 0.2% and at most by about 13.8% than previous works.
Aspect Summarization for Product Reviews based on Attention-based Aspect Extraction
Jun-Nyeong Jeong, Sang-Young Kim, Seong-Tae Kim, Jeong-Jae Lee, Yuchul Jung
http://doi.org/10.5626/JOK.2021.48.12.1318
Recently, document summaries such as articles and papers through machine learning and summary-related research on online reviews are active. In this study, unlike the existing simply summarizing content, a technique was developed for generating an aspect summary by considering various aspects of product reviews. By refining the earphone product review data crawled to build the learning data, 40,000 reviews were obtained. Moreover, we manually constructed 4,000 aspect summaries to be used for our training and evaluation tasks. In particular, we proposed a model that could summarize aspects using only text data using the aspect-based word expansion technique (ABAE). To judge the effectiveness of the proposed technique, we performed experiments according to the use of words related to aspects and the masking ratio during learning. As a result, it was confirmed that the model that randomly masked 25% of the words related to the aspect showed the highest performance, and during verification, the ROUGE was 0.696, and the BERTScore was 0.879.
RDID-GAN: Reconstructing a De-identified Image Dataset to Generate Effective Learning Data
Wonseok Oh, Kangmin Bae, Yuseok Bae
http://doi.org/10.5626/JOK.2021.48.12.1329
Recently, CCTVs have been installed to prevent or handle various social problems, and there are many efforts to develop visual surveillance systems based on deep neural networks. However, the datasets collected from CCTVs are inappropriate to train models due to privacy issues. Therefore, in this paper, we proposed RDID-GAN, an effective dataset de-identification method that can remove privacy issues and negative effects raised by modifying the dataset using a de-identification procedure. RDID-GAN focuses on a de-identified region to produce competitive results by adopting the attention module. Through the experiments, we compared RDID-GAN and the conventional image-to-image translation models qualitatively and quantitatively.
Improvement of Deep Learning Models to Predict the Knowledge Level of Learners based on the EdNet Data
Seulgi Choi, Youngpyo Kim, Sojung Hwang, Heeyoul Choi
http://doi.org/10.5626/JOK.2021.48.12.1335
As online education increases, the field of AI in Education (AIEd), where artificial intelligence is used for education, is being actively studied. Knowledge Tracing (KT), which predicts a student"s knowledge level based on each student"s learning record, is a basic task in the AIEd field. However, there is a lack of utilization of the dataset and research on the KT model architecture. In this paper, we propose to use a total of 11 features, after trying various features related to the problems, and present a new model based on the self-attention mechanism with new query, key, and values, Self-Attentive Knowledge Tracking Extended (SANTE). In experiments, we confirm that the proposed method with the selected features outperforms the previous KT models in terms of the AUC value.
Semi-Supervised Learning Exploiting Robust Loss Function for Sparse Labeled Data
http://doi.org/10.5626/JOK.2021.48.12.1343
This paper proposes a semi-supervised learning method which uses data augmentation and robust loss function when labeled data are extremely sparse. Existing semi-supervised learning methods augment unlabeled data and use one-hot vector labels predicted by the current model if the confidence of the prediction is high. Since it does not use low-confidence data, a recent work has used low-confidence data in the training by utilizing robust loss function. Meanwhile, if labeled data are extremely sparse, the prediction can be incorrect even if the confidence is high. In this paper, we propose a method to improve the performance of a classification model when labeled data are extremely sparse by using predicted probability, instead of one hot vector as the label. Experiments show that the proposed method improves the performance of a classification model.
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