• Journal of Convergence for Information Technology
• /
• v.11 no.7
• /
• pp.31-38
• /
• 2021

In this paper, in order to obtain the optimization of the RNN model used for sentiment analysis, the correlation of each model was studied by observing the trend of loss and accuracy according to hyperparameter tuning. As a research method, after configuring the hidden layer with LSTM and the embedding layer that are most optimized to process sequential data, the loss and accuracy of each model were measured by tuning the unit, batch-size, and embedding size of the LSTM. As a result of the measurement, the loss was 41.9% and the accuracy was 11.4%, and the trend of the optimization model showed a consistently stable graph, confirming that the tuning of the hyperparameter had a profound effect on the model. In addition, it was confirmed that the decision of the embedding size among the three hyperparameters had the greatest influence on the model. In the future, this research will be continued, and research on an algorithm that allows the model to directly find the optimal hyperparameter will continue.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.9
• /
• pp.3298-3321
• /
• 2021

A lot of data hiding schemes have been proposed to embed secret data in the plain cover images or compressed images of various formats, including JPEG, AMBTC, VQ, etc. In this paper, we propose a production process of mosaic images based on three regular images of coffee beans. A primary image is first mimicked by the process to produce a mosaic cover image. A two-layer steganography is applied to hide secret data in the mosaic image. Based on the low visual quality of the mosaic cover image, its PSNR value can be improved about 1.5 dB after embedding 3 bpp. This is achieved by leveraging the newly proposed polarized search mask and the concepts of strong embedding and weak embedding. Applying steganography to the mosaic cover images is a completely new idea and it is promising.

• Journal of Multimedia Information System
• /
• v.9 no.2
• /
• pp.93-102
• /
• 2022

Dependency parsing is a decision problem of the syntactic relation between words in a sentence. Recently, deep learning models are used for dependency parsing based on the word representations in a continuous vector space. However, it causes a mislabeled tagging problem for the proper nouns that rarely appear in the training corpus because it is difficult to express out-of-vocabulary (OOV) words in a continuous vector space. To solve the OOV problem in dependency parsing, we explored the proper noun embedding method according to the embedding unit. Before representing words in a continuous vector space, we replace the proper nouns with a special token and train them for the contextual features by using the multi-layer bidirectional LSTM. Two models of the syllable-based and morpheme-based unit are proposed for proper noun embedding and the performance of the dependency parsing is more improved in the ensemble model than each syllable and morpheme embedding model. The experimental results showed that our ensemble model improved 1.69%p in UAS and 2.17%p in LAS than the same arc-eager approach-based Malt parser.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.6
• /
• pp.1620-1634
• /
• 2023

This paper studies the task of embedding nodes with multiple graphs representing multiple information channels, which is useful in a large volume of network clustering tasks. By learning a node using multiple graphs, various characteristics of the node can be represented and embedded stably. Existing studies using multi-channel networks have been conducted by integrating heterogeneous graphs or limiting common nodes appearing in multiple graphs to have similar embeddings. Although these methods effectively represent nodes, it also has limitations by assuming that all networks provide the same amount of information. This paper proposes a method to overcome these limitations; The proposed method gives different weights according to the source graph when embedding nodes; the characteristics of the graph with more important information can be reflected more in the node. To this end, a novel method incorporating a multi-channel gate layer is proposed to weigh more important channels and ignore unnecessary data to embed a node with multiple graphs. Empirical experiments demonstrate the effectiveness of the proposed multi-channel-based embedding methods.

• Journal of Information Processing Systems
• /
• v.18 no.5
• /
• pp.614-627
• /
• 2022

Aspect-based sentiment analysis is to discover the sentiment polarity towards an aspect from user-generated natural language. So far, most of the methods only use the implicit position information of the aspect in the context, instead of directly utilizing the position relationship between the aspect and the sentiment terms. In fact, neighboring words of the aspect terms should be given more attention than other words in the context. This paper studies the influence of different position embedding methods on the sentimental polarities of given aspects, and proposes a position embedding interactive attention network based on a long short-term memory network. Firstly, it uses the position information of the context simultaneously in the input layer and the attention layer. Secondly, it mines the importance of different context words for the aspect with the interactive attention mechanism. Finally, it generates a valid representation of the aspect and the context for sentiment classification. The model which has been posed was evaluated on the datasets of the Semantic Evaluation 2014. Compared with other baseline models, the accuracy of our model increases by about 2% on the restaurant dataset and 1% on the laptop dataset.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.497-504
• /
• 2020

Constructing speaker embeddings in speaker verification is an important issue. In general, a self-attention mechanism has been applied for speaker embedding encoding. Previous studies focused on training the self-attention in a high-level layer, such as the last pooling layer. In this case, the effect of low-level layers is not well represented in the speaker embedding encoding. In this study, we propose Masked Cross Self-Attentive Encoding (MCSAE) using ResNet. It focuses on training the features of both high-level and low-level layers. Based on multi-layer aggregation, the output features of each residual layer are used for the MCSAE. In the MCSAE, the interdependence of each input features is trained by cross self-attention module. A random masking regularization module is also applied to prevent overfitting problem. The MCSAE enhances the weight of frames representing the speaker information. Then, the output features are concatenated and encoded in the speaker embedding. Therefore, a more informative speaker embedding is encoded by using the MCSAE. The experimental results showed an equal error rate of 2.63 % using the VoxCeleb1 evaluation dataset. It improved performance compared with the previous self-attentive encoding and state-of-the-art methods.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.10a
• /
• pp.949-952
• /
• 2009

In this paper, a watermarking method based on the trellis code with multi-layer is proposed. An image is divided $8{\times}8$ block with no overlapping, and compute the discrete cosine transform(DCT) of each block, and the 12 medium-frequency AC terms from each block are extracted. Next it is compared with gaussian random vectors with zero mean and unit variance. As these processing, the embedding vectors with minimum linear correlation can be obtained by Viterbi algorithm at each layer of trellis coding. To evaluate the performance of proposed method, the average bit error rate of watermark message is calculated from different several images.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.2
• /
• pp.148-154
• /
• 2021

Text-Independent speaker verification needs to extract text-independent speaker embedding to improve generalization performance. However, deep neural networks that depend on training data have the potential to overfit text information instead of learning the speaker information when repeatedly learning from the identical time series. In this paper, to prevent the overfitting, we propose a segment unit shuffling layer that divides and rearranges the input layer or a hidden layer along the time axis, thus mixes the time series information. Since the segment unit shuffling layer can be applied not only to the input layer but also to the hidden layers, it can be used as generalization technique in the hidden layer, which is known to be effective compared to the generalization technique in the input layer, and can be applied simultaneously with data augmentation. In addition, the degree of distortion can be adjusted by adjusting the unit size of the segment. We observe that the performance of text-independent speaker verification is improved compared to the baseline when the proposed segment unit shuffling layer is applied.

• Journal of Internet of Things and Convergence
• /
• v.8 no.5
• /
• pp.17-25
• /
• 2022

Recently the massive amount of data has been generated because of the number of edge devices increases. And especially, the number of raw unstructured HTML documents has been increased. Therefore, MRC(Machine Reading Comprehension) in which a natural language processing model finds the important information within an HTML document is becoming more important. In this paper, we propose HTDE(HTML Tag Depth Embedding Method), which allows the BERT to train the depth of the HTML document structure. HTDE makes a tag stack from the HTML document for each input token in the BERT and then extracts the depth information. After that, we add a HTML embedding layer that takes the depth of the token as input to the step of input embedding of BERT. Since tokenization using HTDE identifies the HTML document structures through the relationship of surrounding tokens, HTDE improves the accuracy of BERT for HTML documents. Finally, we demonstrated that the proposed idea showing the higher accuracy compared than the accuracy using the conventional embedding of BERT.

• The Journal of Korean Medicine Ophthalmology and Otolaryngology and Dermatology
• /
• v.36 no.4
• /
• pp.113-121
• /
• 2023

Objectives : The purpose of this study is to analyze the treatment mechanism of Thread-embedding acupuncture, which is used in Korean medicine beauty treatment. Methods : After searching papers published up to January 1, 2023 using the keyword "Thread-embedding" through the OASIS site, we selected beauty papers that mentioned the treatment mechanism of Thread-embedding acupuncture. Results : A total of 60 papers were retrieved: 19 papers on the topic of cosmetic diseases, 35 papers on the theme of other diseases, and 6 papers written unrelated to diseases. Among the 19 papers on the topic of cosmetic diseases, one unreadable paper was excluded. Among the 18 papers, we finally selected 6 papers that mentioned treatment mechanisms: 2 on facial wrinkles, 2 on obesity, 1 on breast enlargement, and 1 on transdermal hydration. The treatment mechanism of Thread-embedding acupuncture is that in the case of facial wrinkles, polydioxanone(PDO) is embedded to fill the volume, and as it decomposes, it causes a tissue reaction around the area. In obesity, it promotes fat decomposition by improving circulation, and promotes breast enlargement and elasticity through collagen formation. In transdermal hydration, it induces the production of surrounding fibers to increase skin elasticity and moisture. Conclusions : Thread-embedding acupuncture appears to have a cosmetic effect through a mechanism that promotes the production of collagen and elastic fibers in the subepidermal dermal layer and increases the activity of skin moisturizing factors during the absorption process after the PDO suture is embedded.