• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.10a
• /
• pp.351-354
• /
• 2016

The competition of Go between AlphaGo and Lee Sedol attracted global interest leading AlphaGo to victory. The core function of AlphaGo is deep-learning system, studying by computer itself. Afterwards, the utilization of deep-learning system using artificial intelligence is said to be verified. Recently, the government passed the loT Act and developing its business model to promote loT. This study is on analyzing IoT business environment using deep-learning AI and constructing specialized business models.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.272-273
• /
• 2019

VVC(Versatile Video Coding)는 YUV 입력 영상에 대하여 Luma 성분과 Chroma 성분에 대하여 각각 다른 최적의 방법으로 블록분할 후 해당 블록에 대해서 화면 내 예측 또는 화면 간 예측을 수행하고, 예측영상과 원본영상의 차이를 변환, 양자화하여 압축한다. 이 과정에서 복원영상에는 블록화 노이즈, 링잉 노이즈, 블러링 노이즈 발생한다. 본 논문에서는 인코더에서 원본영상과 복원영상의 잔차신호에 대한 MAE(Mean Absolute Error)를 추가정보로 전송하여 이 추가정보와 복원영상을 이용하여 Deep Learning 기반의 신경망 네트워크로 영상의 품질을 높이는 방법을 제안한다. 복원영상의 노이즈를 감소시키기 위하여 영상을 $32{\times}32$블록의 임의로 분할하고, DenseNet기반의 UNet 구조로 네트워크를 구성하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2019.05a
• /
• pp.347-348
• /
• 2019

In this paper, we propose a method to increase the resolution of a digital hologram by using deep learning. We reduced the size of holograms for training super-resolution algorithm and created a dataset using a subset of them. We trained the network model with the generated dataset and confirmed the PSNR over 31dB.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.11
• /
• pp.4557-4572
• /
• 2020

The Alaska competition provides an opportunity to study the practical problems of real-world steganalysis. Participants are required to solve steganalysis involving various embedding schemes, inconsistency JPEG Quality Factor and various processing pipelines. In this paper, we propose a method to ensemble multiple deep learning steganalyzers. We select SRNet and RESDET as our base models. Then we design a three-layers model ensemble network to fuse these base models and output the final prediction. By separating the three colors channels for base model training and feature replacement strategy instead of simply merging features, the performance of the model ensemble is greatly improved. The proposed method won second place in the Alaska 1 competition in the end.

• Journal of Korea Multimedia Society
• /
• v.24 no.5
• /
• pp.651-658
• /
• 2021

In this paper, we propose methods for character recognition and classification for table items through deep learning. First, table areas are detected in a document image through CNN. After that, table areas are separated by separators such as vertical lines. The text in document is recognized through a neural network combined with CNN and RNN. To correct errors in the character recognition, multiple candidates for the recognized result are provided for a sentence which has low recognition accuracy.

• Journal of the Semiconductor & Display Technology
• /
• v.20 no.2
• /
• pp.103-108
• /
• 2021

The TCD is used as one of the indicators for determining whether TSV Hole is defective. If the TCD is not normal size, it can lead to contamination of the CMP equipment or failure to connect the upper and lower chips. We propose a deep learning model for measuring the TCD. To verify the performance of the proposed model, we compared the prediction results of the proposed model for 2461 via holes with the CD-SEM measurement data and the prediction results of the existing model. Although the number of trainable parameters in the proposed model was about one two-thousandth of the existing model, the results were comparable. The experiment showed that the correlation between CD-SEM and the prediction results of the proposed model measured 98%, the mean absolute difference was 0.051um, the standard deviation of the absolute difference was 0.045um, and the maximum absolute difference was 0.299um on average.

• Journal of information and communication convergence engineering
• /
• v.19 no.1
• /
• pp.48-53
• /
• 2021

Herein, we propose a document analysis system that analyzes papers or reports transformed into XML(Extensible Markup Language) format. It reads the document specified by the user, extracts keywords from the document, and compares the frequency of keywords to extract the top-three keywords. It maintains the order of the paragraphs containing the keywords and removes duplicated paragraphs. The frequency of the top-three keywords in the extracted paragraphs is re-verified, and the paragraphs are partitioned into 10 sections. Subsequently, the importance of the relevant areas is calculated and compared. By notifying the user of areas with the highest frequency and areas with higher importance than the average frequency, the user can read only the main content without reading all the contents. In addition, the number of paragraphs extracted through the deep learning model and the number of paragraphs in a section of high importance are predicted.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.5
• /
• pp.670-673
• /
• 2020

Recent advances in artificial intelligence have allowed for easier sentiment analysis (e.g. positive or negative forecast) of documents such as a finance reports. In this paper, we investigate a method to apply text mining techniques to extract in the financial report using deep learning, and propose an accounting model for the effects of sentiment values in financial information. For sentiment analysis with keyword detection in the financial report, we suggest the input layer with extracted keywords, hidden layers by learned weights, and the output layer in terms of sentiment scores. Our approaches can help more effective strategy for potential investors as a professional guideline using sentiment values.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.12 no.1
• /
• pp.768-783
• /
• 2020

Typically, a Dynamic Positioning System (DPS) uses a PID feed-back system, and it often adopts a wind feed-forward system because of its easier implementation than a feed-forward system based on current or wave. But, because a ship's drifting motion is caused by wind, current, and wave drift loads, all three environmental loads should be considered. In this study, a motion predictive control for the PID feedback system of the DPS is proposed, which considers the three environmental loads by utilizing predicted drifted ship positions in the future since it contains information about the three environmental loads from the moment to the future. The prediction accuracy for the future drifted ship position is ensured by adopting deep learning algorithms and a replay buffer. Finally, it is shown that the proposed motion predictive system results in better station-keeping performance than the wind feed-forward system.

• Journal of information and communication convergence engineering
• /
• v.19 no.2
• /
• pp.73-78
• /
• 2021

As interest in beauty has increased, various studies have been conducted, and related companies have considered the anthropometric data handled between humans and interfaces as an important factor. However, owing to the nature of 3D human body scanners used to extract anthropometric data, it is difficult to accurately analyze a user's body shape until a service is provided because the user only scans and extracts data. To solve this problem, the body shape of several users was analyzed, and the collected anthropometric data were obtained using a 3D human body scanner. After processing the extracted data and the anthropometric data, a custom deep learning model was designed, the designed model was learned, and the user's body shape information was predicted to provide a service suitable for the body shape. Through this approach, it is expected that the user's body shape information can be predicted using a 3D human body scanner, based upon which a beauty service can be provide.