• 전기학회논문지
• /
• 제67권12호
• /
• pp.1665-1670
• /
• 2018

A convolutional neural network (CNN), which is one of the deep learning models, has been very successful in a variety of computer vision tasks. Filters of a CNN are automatically generated, however, they can be further optimized since there exist the possibility of existing redundant and less important features. Therefore, the aim of this paper is a filter reduction to accelerate and compress CNN models. Evolutionary algorithms is adopted to remove the unnecessary filters in order to minimize the parameters of CNN networks while maintaining a good performance of classification. We demonstrate the proposed filter reduction methods performing experiments on CIFAR10 data based on the classification performance. The comparison for three approaches is analysed and the outlook for the potential next steps is suggested.

• International Journal of Internet, Broadcasting and Communication
• /
• 제13권2호
• /
• pp.208-217
• /
• 2021

3D-CNN is one of the deep learning techniques for learning time series data. However, these three-dimensional learning can generate many parameters, requiring high performance or having a significant impact on learning speed. We will use these 3D-CNNs to learn hand gesture and find the parameters that showed the highest accuracy, and then analyze how the accuracy of 3D-CNN varies through input data changes without any structural changes in 3D-CNN. First, choose the interval of the input data. This adjusts the ratio of the stop interval to the gesture interval. Secondly, the corresponding interframe mean value is obtained by measuring and normalizing the similarity of images through interclass 2D cross correlation analysis. This experiment demonstrates that changes in input data affect learning accuracy without structural changes in 3D-CNN. In this paper, we proposed two methods for changing input data. Experimental results show that input data can affect the accuracy of the model.

• 반도체디스플레이기술학회지
• /
• 제19권4호
• /
• pp.65-70
• /
• 2020

CNN's technologies that represent emotional detection include primitive CNN algorithms, deployment normalization, and drop-off. We present the methods and data of the three experiments in this paper. The training database and the test database are set up differently. The first experiment is to extract emotions using Batch Normalization, which complemented the shortcomings of distribution. The second experiment is to extract emotions using Dropout, which is used for rapid computation. The third experiment uses CNN using convolution and maxpooling. All three results show a low detection rate, To supplement these problems, We will develop a deep learning algorithm using feature extraction method specialized in image processing field.

• 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.

• 통합자연과학논문집
• /
• 제13권4호
• /
• pp.147-151
• /
• 2020

In this paper, we propose a method to detect concept drift by applying Convolutional Neural Network (CNN) in a data stream environment. Since the conventional method compares only the final output value of the CNN and detects it as a concept drift if there is a difference, there is a problem in that the actual input value of the data stream reacts sensitively even if there is no significant difference and is incorrectly detected as a concept drift. Therefore, in this paper, in order to reduce such errors, not only the output value of CNN but also the probability vector are used. First, the data entered into the data stream is patterned to learn from the neural network model, and the difference between the output value and probability vector of the current data and the historical data of these learned neural network models is compared to detect the concept drift. The proposed method confirmed that only CNN output values could be used to reduce detection errors compared to how concept drift were detected.

• 반도체디스플레이기술학회지
• /
• 제19권4호
• /
• pp.71-76
• /
• 2020

In this paper, we propose a super-resolution method that reconstructs compressed low-resolution images into high-resolution images. We propose a CNN model with a small number of parameters, and even if quantization is applied to the proposed model, super-resolution can be implemented without deteriorating the image quality. To further improve the quality of the compressed low-resolution image, a new degradation model was proposed instead of the existing bicubic degradation model. The proposed degradation model is used only in the training process and can be applied by changing only the parameter values to the original CNN model. In the super-resolution image applying the proposed degradation model, visual artifacts caused by image compression were effectively removed. As a result, our proposed method generates higher PSNR values at compressed images and shows better visual quality, compared to conventional CNN-based SR methods.

• 전력전자학회논문지
• /
• 제27권1호
• /
• pp.48-55
• /
• 2022

This paper proposes a battery remaining useful life (RUL) prediction method using a deep learning-based EMD-CNN-LSTM hybrid method. The proposed method pre-processes capacity data by applying empirical mode decomposition (EMD) and predicts the remaining useful life using CNN-LSTM. CNN-LSTM is a hybrid method that combines convolution neural network (CNN), which analyzes spatial features, and long short term memory (LSTM), which is a deep learning technique that processes time series data analysis. The performance of the proposed remaining useful life prediction method is verified using the battery aging experiment data provided by the NASA Ames Prognostics Center of Excellence and shows higher accuracy than does the conventional method.

• International Journal of Advanced Culture Technology
• /
• 제11권3호
• /
• pp.297-301
• /
• 2023

The study identified the various causes of odor problems, the discomfort they cause, and the importance of the public health and environmental issues associated with them. To solve the odor problem, you must identify the cause and perform an accurate analysis. Therefore, we proposed a CNN-Transformer hybrid model (CTHM) that combines CNN and Transformer and evaluated its performance. It was evaluated using a dataset consisting of 120,000 odor samples, and experimental results showed that CTHM achieved an accuracy of 93.000%, a precision of 92.553%, a recall of 94.167%, an F1 score of 92.880%, and an RMSE of 0.276. Our results showed that CTHM was suitable for odor analysis and had excellent prediction performance. Utilization of this model is expected to help address odor problems and alleviate public health and environmental concerns.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2022년도 춘계학술발표대회
• /
• pp.494-496
• /
• 2022

엣지 디바이스에서 딥러닝을 활용하기 위하여 CNN 경량화 연구들이 진행되고 있다. 경량 CNN 은 대부분 고정 소수점을 사용하며, 계층에 따라 정밀도는 달라진다. 본 논문에서는 경량 CNN 을 지원하기 위하여, 사용 계층에 따라 정밀도를 선택할 수 있는 가변 정밀도 곱셈기를 제안한다. 제안하는 가변 정밀도 곱셈기는 낮은 정밀도 곱셈기를 병합하는 구조로, 정밀도가 낮을 때는 병렬 처리를 통해 효율을 높인다. 제안하는 곱셈기를 Verilog HDL로 설계하고 ModelSim 에서 동작을 확인하였다. 설계된 곱셈기는 계층별로 정밀도가 다른 CNN 가속기에서 효율적으로 적용될 것으로 기대된다.

• Journal of information and communication convergence engineering
• /
• 제21권2호
• /
• pp.159-166
• /
• 2023

The COI gene is a sequence of approximately 650 bp at the 5' terminal of the mitochondrial Cytochrome c Oxidase subunit I (COI) gene. As an effective DeoxyriboNucleic Acid (DNA) barcode, it is widely used for the taxonomic identification and evolutionary analysis of species. We created a CNN-LSTM hybrid model by combining the gene features partially extracted by the Long Short-Term Memory ( LSTM ) network with the feature maps obtained by the CNN. Compared to K-Means Clustering, Support Vector Machines (SVM), and a single CNN classification model, after training 278 samples in a training set that included 15 genera from two orders, the CNN-LSTM hybrid model achieved 94% accuracy in the test set, which contained 118 samples. We augmented the training set samples and four genera into four orders, and the classification accuracy of the test set reached 100%. This study also proposes calculating the cosine similarity between the training and test sets to initially assess the reliability of the predicted results and discover new species.