• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.280-282
• /
• 2020

최근 사물인터넷(IoT), 자율주행과 같이 기계 간의 통신이 요구되는 서비스가 늘어감에 따라, 기계 임무 수행에 최적화된 데이터의 생성 및 압축에 대한 필요성이 증가하고 있다. 또한, 사물인터넷과 인공지능(AI)이 접목된 기술이 주목을 받으면서 딥러닝 모델에서 추출되는 특징(feature)을 디바이스에서 클라우드로 전송하는 방안에 관한 연구가 진행되고 있으며, 국제 표준화 기구인 MPEG에서는 '기계를 위한 부호화(Video Coding for Machine: VCM)'에 대한 표준 기술 개발을 진행 중이다. 딥러닝으로 특징을 추출하는 가장 대표적인 방법으로는 합성곱 신경망(Convolutional Neural Network: CNN)이 있으며, 오토인코더는 입력층과 출력층의 구조를 동일하게 하여 출력을 가능한 한 입력에 근사시키고 은닉층을 입력층보다 작게 구성하여 차원을 축소함으로써 데이터를 압축하는 딥러닝 기반 이미지 압축 방식이다. 이에 본 논문에서는 이러한 오토인코더의 성질을 이용하여 CNN 기반의 이미지 분류 네트워크의 합성곱 신경망으로부터 추출된 feature에 오토인코더를 적용하여 압축하는 방안을 제안한다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.01a
• /
• pp.69-70
• /
• 2021

본 논문은 PPM으로 관리되고 있는 자동차 부품 제조 공정에서 검사자의 육안검사 방법을 대체하기 위해 머신비전 및 CNN 기반 불량 검출 시스템으로 제안되었던 방식들의 단점을 개선하기 위하여 기존 머신 비전 기술에 합성곱 오토인코더 모델을 적용하여 단점을 해결하였다. 본 논문에서 제시한 오토인코더를 이용하는 방법은 정상 생산품의 이미지만으로 학습을 진행하고, 학습된 모델은 불량 부위가 포함된 이미지를 입력받아 정상 이미지로 출력한다. 이 방법을 사용하여 불량의 부위와 크기를 알 수 있었으며 불량 여부의 판단은 임계치에 의한 불량 부위의 화소 수 계산으로 판단하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.4
• /
• pp.337-346
• /
• 2021

Detection of abnormal signal generally can be done by using features of normal signals as main information because of data imbalance. This paper propose an efficient method for abnormal signal detection using parallel AutoEncoder (AE) which can use features of abnormal signals as well. The proposed Parallel AE (PAE) is composed of a normal and an abnormal reconstructors having identical AE structure and train features of normal and abnormal signals, respectively. The PAE can effectively solve the imbalanced data problem by sequentially training normal and abnormal data. For further detection performance improvement, additional binary classifier can be added to the PAE. Through experiments using public acoustic data, we obtain that the proposed PAE shows Area Under Curve (AUC) improvement of minimum 22 % at the expenses of training time increased by 1.31 ~ 1.61 times to the single AE. Furthermore, the PAE shows 93 % AUC improvement in detecting abnormal underwater acoustic signal when pre-trained PAE is transferred to train open underwater acoustic data.

• The Korean Journal of Applied Statistics
• /
• v.35 no.1
• /
• pp.49-62
• /
• 2022

Realized volatility is well known to have long memory, strong association with other global financial markets and interdependences among macroeconomic indices such as exchange rate, oil price and interest rates. This paper proposes autoencoder factor-augmented heterogeneous autoregressive (AE-FAHAR) model for realized volatility forecasting. AE-FAHAR incorporates long memory using HAR structure, and exogenous variables into few factors summarized by autoencoder. Autoencoder requires intensive calculation due to its nonlinear structure, however, it is more suitable to summarize complex, possibly nonstationary high-dimensional time series. Our AE-FAHAR model is shown to have smaller out-of-sample forecasting error in empirical analysis. We also discuss pre-training, ensemble in autoencoder to reduce computational cost and estimation errors.

• Journal of Internet Computing and Services
• /
• v.22 no.1
• /
• pp.23-30
• /
• 2021

This paper proposes a noise-tolerant image classification system using multiple autoencoders. The development of deep learning technology has dramatically improved the performance of image classifiers. However, if the images are contaminated by noise, the performance degrades rapidly. Noise added to the image is inevitably generated in the process of obtaining and transmitting the image. Therefore, in order to use the classifier in a real environment, we have to deal with the noise. On the other hand, the autoencoder is an artificial neural network model that is trained to have similar input and output values. If the input data is similar to the training data, the error between the input data and output data of the autoencoder will be small. However, if the input data is not similar to the training data, the error will be large. The proposed system uses the relationship between the input data and the output data of the autoencoder, and it has two phases to classify the images. In the first phase, the classes with the highest likelihood of classification are selected and subject to the procedure again in the second phase. For the performance analysis of the proposed system, classification accuracy was tested on a Gaussian noise-contaminated MNIST dataset. As a result of the experiment, it was confirmed that the proposed system in the noisy environment has higher accuracy than the CNN-based classification technique.

• Journal of the Korea Computer Graphics Society
• /
• v.27 no.5
• /
• pp.33-43
• /
• 2021

In this paper, we propose a framework that transfers the information of style motions to content motions based on a variational autoencoder network combined with a style encoding in the latent space. Because we transfer a style to a content motion that is sampled from a variational autoencoder, we can increase the diversity of existing motion data. In addition, we can improve the unnatural motions caused by decoding a new latent variable from style transfer. That improvement was achieved by additionally using the velocity information of motions when generating next frames.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.3
• /
• pp.568-577
• /
• 2017

IEEE Std 1609.2 was defined for the support of communication security functions in the WAVE (Wireless Access in Vehicular Environments) system. IEEE Std 1609.2 defined the message structures of the security services and managements on the vehicular communication by using ASN.1 (Abstract Syntax Notation One). Also, this security message structures shall be encoded using the COER (Canonical Octet Encoding Rules). In this paper, we designed and implemented the IEEE Std 1609.2 message encoder/decoder handling the security messages defined in IEEE Std 1609.2. The designed encoder/decoder consists of three modules as follows : a module generating the message of C language data structures in accord with IEEE Std 1609.2 message structures, a message encoder module, a message decoder module. And the encoder/decoder was implemented on the Linux environment. Also we analyzed the performance by measuring the performance speed of the encoder/decoder implemented.

• The Journal of Society for e-Business Studies
• /
• v.26 no.1
• /
• pp.127-140
• /
• 2021

Recently, sequence data containing time information, such as sensor measurement data and purchase history, has been generated in various applications. So far, many methods for finding sequences that are significantly different from other sequences among given sequences have been proposed. However, most of them have a limitation that they consider only the order of elements in the sequences. Therefore, in this paper, we propose a new anomalous sequence detection method that considers both the order of elements and the time interval between elements. The proposed method uses an extended LSTM autoencoder model, which has an additional layer that converts a sequence into a form that can help effectively learn both the order of elements and the time interval between elements. The proposed method learns the features of the given sequences with the extended LSTM autoencoder model, and then detects sequences that the model does not reconstruct well as anomalous sequences. Using experiments on synthetic data that contains both normal and anomalous sequences, we show that the proposed method achieves an accuracy close to 100% compared to the method that uses only the traditional LSTM autoencoder.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.6
• /
• pp.500-510
• /
• 2023

Passive sonar signals mainly contain both normal and abnormal signals. The abnormal signals mixed with normal signals are primarily detected using an AutoEncoder (AE) that learns only normal signals. However, existing AEs may perform inaccurate detection by reconstructing distorted normal signals from mixed signal. To address these limitations, we propose an abnormal signal detection model based on a Recurrent Neural Network (RNN) and vector quantization. The proposed model generates a codebook representing the learned latent vectors and detects abnormal signals more accurately through the proposed search process of code vectors. In experiments using publicly available underwater acoustic data, the AE and Variational AutoEncoder (VAE) using the proposed method showed at least a 2.4 % improvement in the detection performance and at least a 9.2 % improvement in the extraction performance for abnormal signals than the existing models.

• The Journal of the Convergence on Culture Technology
• /
• v.10 no.1
• /
• pp.603-608
• /
• 2024

Nonresponse and missing values are caused by sample dropouts and avoidance of answers to surveys. In this case, problems with the possibility of information loss and biased reasoning arise, and a replacement of missing values with appropriate values is required. In this paper, as an alternative to missing values imputation, we compare several replacement methods, which use mean, linear regression, random forest, K-nearest neighbor, autoencoder and denoising autoencoder based on deep learning. These methods of imputing missing values are explained, and each method is compared by using continuous simulation data and real data. The comparison results confirm that in most cases, the performance of the random forest imputation method and the denoising autoencoder imputation method are better than the others.