• 한국기계기술학회지
• /
• 제20권6호
• /
• pp.776-782
• /
• 2018

After CNN basic structure was introduced by LeCun in 1989, there has not been a major structure change except for more deep network until recently. The deep network enhances the expression power due to improve the abstraction ability of the network, and can learn complex problems by increasing non linearity. However, the learning of a deep network means that it has vanishing gradient or longer learning time. In this study, we proposes a CNN structure with MLP layer. The proposed CNNs are superior to the general CNN in their classification performance. It is confirmed that classification accuracy is high due to include MLP layer which improves non linearity by experiment. In order to increase the performance without making a deep network, it is confirmed that the performance is improved by increasing the non linearity of the network.

• 융합신호처리학회논문지
• /
• 제22권1호
• /
• pp.30-37
• /
• 2021

카메라 영상을 이용한 3차원 객체 추적 기술은 증강현실 응용 분야를 위한 핵심 기술이다. 영상 분류, 객체 검출, 영상 분할과 같은 컴퓨터 비전 작업에서 CNN(Convolutional Neural Network)의 인상적인 성공에 자극 받아, 3D 객체 추적을 위한 최근의 연구는 딥러닝(deep learning)을 활용하는 데 초점을 맞추고 있다. 본 논문은 이러한 딥러닝을 활용한 3차원 객체 추적 방법들을 살펴본다. 딥러닝을 활용한 3차원 객체 추적을 위한 주요 방법들을 설명하고, 향후 연구 방향에 대해 논의한다.

• Journal of information and communication convergence engineering
• /
• 제19권1호
• /
• pp.22-28
• /
• 2021

Convolutional neural networks (CNNs) are one of the most frequently used artificial intelligence techniques. Among CNN-based applications, small and timing-sensitive applications have emerged, which must be reliable to prevent severe accidents. However, as the small and timing-sensitive systems do not have sufficient system resources, they do not possess proper error protection schemes. In this paper, we propose MATE, which is a low-cost CNN weight error correction technique. Based on the observation that all mantissa bits are not closely related to the accuracy, MATE replaces some mantissa bits in the weight with error correction codes. Therefore, MATE can provide high data protection without requiring additional memory space or modifying the memory architecture. The experimental results demonstrate that MATE retains nearly the same accuracy as the ideal error-free case on erroneous DRAM and has approximately 60% accuracy, even with extremely high bit error rates.

• Journal of Information Processing Systems
• /
• 제18권2호
• /
• pp.197-207
• /
• 2022

The applications and user interfaces (UIs) of smart mobile devices are constantly diversifying. For example, deep learning can be an innovative solution to classify widgets in screen images for increasing convenience. To this end, the present research leverages captured images and the ReDraw dataset to write deep learning datasets for image classification purposes. First, as the validation for datasets using ResNet50 and EfficientNet, the experiments show that the dataset composed in this study is helpful for classification according to a widget's functionality. An implementation for widget detection and classification on RetinaNet and EfficientNet is then executed. Finally, the research suggests the Widg-C and Widg-D datasets-a deep learning dataset for identifying the widgets of smart devices-and implementing them for use with representative convolutional neural network models.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권11호
• /
• pp.3658-3679
• /
• 2022

Classification of persons wearing and not wearing face masks in images has emerged as a new computer vision problem during the COVID-19 pandemic. In order to address this problem and scale up the research in this domain, in this paper a hybrid technique by employing ResNet-101 and multi-layer perceptron (MLP) classifier has been proposed. The proposed technique is tested and validated on a self-created face masks classification dataset and a standard dataset. On self-created dataset, the proposed technique achieved a classification accuracy of 97.3%. To embrace the proposed technique, six other state-of-the-art CNN feature extractors with six other classical machine learning classifiers have been tested and compared with the proposed technique. The proposed technique achieved better classification accuracy and 1-6% higher precision, recall, and F1 score as compared to other tested deep feature extractors and machine learning classifiers.

• 대한임베디드공학회논문지
• /
• 제18권3호
• /
• pp.101-108
• /
• 2023

Optimizing the performance of deep neural networks on embedded systems is a challenging task that requires efficient compilers and runtime systems. We propose a TVM-based approach that consists of three steps: quantization, auto-scheduling, and ahead-of-time compilation. Our approach reduces the computational complexity of models without significant loss of accuracy, and generates optimized code for various hardware platforms. We evaluate our approach on three representative CNNs using ImageNet Dataset on the NVIDIA Jetson AGX Xavier board and show that it outperforms baseline methods in terms of processing speed.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.878-879
• /
• 2021

Most state-of-the-art CNNs for action recognition are based on a two-stream architecture: RGB frames stream represents the appearance and the optical flow stream interprets the motion of action. However, the cost of optical flow computation is very high and then it increases action recognition latency. We introduce a design strategy for action recognition inspired by a two-stream network and teacher-student architecture. There are two sub-networks in our neural networks, the optical flow sub-network as a teacher and the RGB frames sub-network as a student. In the training stage, we distill the feature from the teacher as a baseline to train student sub-network. In the test stage, we only use the student so that the latency reduces without computing optical flow. Our experiments show that its advantages over two-stream architecture in both speed and performance.

• 한국해양공학회지
• /
• 제37권5호
• /
• pp.181-189
• /
• 2023

The fuel consumption of marine diesel engines holds paramount importance in contemporary maritime transportation and shapes energy efficiency strategies of ocean-going vessels. Nonetheless, a noticeable gap in knowledge prevails concerning the influence of ship hull conditions and propeller roughness on fuel consumption. This study bridges this gap by utilizing artificial intelligence techniques in Matlab, particularly convolutional neural networks (CNNs) to comprehensively investigate these factors. We propose a time-series prediction model that was built on numerical simulations and aimed at forecasting ship hull and propeller conditions. The model's accuracy was validated through a meticulous comparison of predictions with actual ship-hull and propeller conditions. Furthermore, we executed a comparative analysis juxtaposing predictive outcomes with navigational environmental factors encompassing wind speed, wave height, and ship loading conditions by the fuzzy clustering method. This research's significance lies in its pivotal role as a foundation for fostering a more intricate understanding of energy consumption within the realm of maritime transport.

• International Journal of Advanced Culture Technology
• /
• 제11권4호
• /
• pp.464-469
• /
• 2023

As modern society develops, the number of diseases caused by bad smells is increasing. As it can harm people's health, it is important to predict in advance the extent to which bad smells may occur, inform the public about this, and take preventive measures. In this paper, we propose a hybrid neural network structure of CNN and LSTM that can be used to detect or predict the occurrence of odors, which are most required in manufacturing or real life, using odor complex sensors. In addition, the proposed learning model uses a complex odor sensor to receive four types of data, including hydrogen sulfide, ammonia, benzene, and toluene, in real time, and applies this data to the inference model to detect and predict the odor state. The proposed model evaluated the prediction accuracy of the training model through performance indicators based on accuracy, and the evaluation results showed an average performance of more than 94%.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2024년도 춘계학술발표대회
• /
• pp.404-407
• /
• 2024

동형암호는 주목받는 차세대 프라이버시 보존 기술이며, 이를 활용한 신경망 연산 연구들이 많이 수행되고 있다. 여러 체계들 중 TFHE는 비산술 연산을 직접 연산할 수 있으나 다른 체계들보다 매우 느리다는 단점이 있다. 본 연구는 정확도 하락을 최소화하며 성능 개선을 통해 다른 체계인 CKKS보다 빠른 TFHE기반 CNN 연산이 가능하도록 하는 TFHE 비산술 연산의 양자화 기술을 소개한다.