• The Journal of the Acoustical Society of Korea
• /
• v.43 no.2
• /
• pp.243-252
• /
• 2024

Among the Foley sound generation models that have recently begun to be studied, a sound generation technique using the Vector Quantized-Variational AutoEncoder (VQ-VAE) structure and generation model such as Pixelsnail are one of the important research subjects. On the other hand, in the field of deep learning-based acoustic signal compression, residual vector quantization technology is reported to be more suitable than the conventional VQ-VAE structure. Therefore, in this paper, we aim to study whether residual vector quantization technology can be effectively applied to the Foley sound generation. In order to tackle the problem, this paper applies the residual vector quantization technique to the conventional VQ-VAE-based Foley sound generation model, and in particular, derives a model that is compatible with the existing models such as Pixelsnail and does not increase computational resource consumption. In order to evaluate the model, an experiment was conducted using DCASE2023 Task7 data. The results show that the proposed model enhances about 0.3 of the Fréchet audio distance. Unfortunately, the performance enhancement was limited, which is believed to be due to the decrease in the resolution of time-frequency domains in order to do not increase consumption of the computational resources.

• Journal of the Korean housing association
• /
• v.28 no.2
• /
• pp.13-21
• /
• 2017

A well designed disabled home needs to consider various design principles and variables. It should consider not only physical accessibility but also psychological aspects. Previous research studies have shown that barrier-free/universal designs which were primarily focused on physical access and usage of facilities and building operations. This research paper will examine, a selection of refurbished disabled homes, introduced by the Korean Disabled People's Development Institutes in 2013. The plan samples are to be analyzed using the Visual Access and Exposure spatial analysis program coupled with supporting information extracted from consultations and feedback from experienced professional disabled home staff members. This research paper aims to propose the usage and viability of VAE Analysis in the design and planning of disabled home layouts. The purpose of this study is to specify the differences in visual spatial relationships between the plans before and after refurbishment in accordance to staff and user requirements. This will ensure a bettered environment for the users and ensuring an optimized of spatial programming and building operation and usage.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.531-534
• /
• 2022

승차 공유, 카풀, 렌터카의 이용률이 증가하면서 많은 사용자가 동일한 차량에 로컬 액세스 할 수 있는 시나리오가 더욱 보편화됨에 따라 차량 네트워크에 대한 공격 가능성이 커지고 있다. 차량용 CAN Bus Network에 대한 DoS(Denial of Service), Fuzzy Attack 및 Replay Attack과 같은 공격은 일부 ECU(Electronic Controller Unit) 비활성 및 작동 불능 상태를 유발한다. 에어백, 제동 시스템과 같은 필수 시스템이 작동 불가 상태가 되어 운전자에게 치명적인 결과를 초래할 수 있다. 차량 네트워크 침입 탐지를 위하여 많은 연구가 진행되고 있으나, 기존 화이트리스트를 이용한 탐지 방법은 새로운 유형의 공격이 발생하거나 희소성이 높은 공격일 때 탐지하기 어렵다. 본 논문에서는 인공신경망 기반의 CAN 버스 네트워크 침입 탐지 기법을 제안한다. 제안하는 침입 탐지 기법은 2단계로 나누어 진다. 1단계에서 정상 패킷 분포를 학습한 VAE 모형이 이상 탐지를 수행한다. 이상 패킷으로 판정될 경우, 2단계에서 인코더로부터 추출된 잠재변수와 VAE의 재구성 오차를 이용하여 공격 유형을 분류한다. 분류 결과의 신뢰점수(Confidence score)가 임계치보다 낮을 경우 학습하지 않은 공격으로 판단한다. 본 연구 결과물은 정보보호 연구·개발 데이터 첼린지 2019 대회의 차량 이상징후 탐지 트랙에서 제공하는 정상 및 3종의 차량 공격시도 패킷 데이터를 대상으로 성능을 평가하였다. 실험을 통해 자동차 제조사의 규칙이나 정책을 사전에 정의하지 않더라도 낮은 오탐율로 비정상 패킷을 탐지해 낼 수 있음을 확인할 수 있다.

• Journal of Digital Convergence
• /
• v.15 no.12
• /
• pp.341-347
• /
• 2017

As the amount of document data increases, the user needs summarized information to understand the document. However, existing document summary research methods rely on overly simple statistics, so there is insufficient research on multiple document summaries for ambiguity of sentences and meaningful sentence generation. In this paper, we investigate semantic connection and preprocessing process to process unnecessary information. Based on the vocabulary semantic pattern information, we propose a multi-document summarization method that enhances semantic connectivity between sentences using VAE. Using sentence word vectors, we reconstruct sentences after learning from compressed information and attribute discriminators generated as latent variables, and semantic connection processing generates a natural summary sentence. Comparing the proposed method with other document summarization methods showed a fine but improved performance, which proved that semantic sentence generation and connectivity can be increased. In the future, we will study how to extend semantic connections by experimenting with various attribute settings.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.6
• /
• pp.578-586
• /
• 2021

We propose a semi-supervised learning method based on Variational AutoEncoder (VAE) and Unsupervised Data Augmentation (UDA) to improve the performance of an end-to-end speech recognizer. In the proposed method, first, the VAE-based augmentation model and the baseline end-to-end speech recognizer are trained using the original speech data. Then, the baseline end-to-end speech recognizer is trained again using data augmented from the learned augmentation model. Finally, the learned augmentation model and end-to-end speech recognizer are re-learned using the UDA-based semi-supervised learning method. As a result of the computer simulation, the augmentation model is shown to improve the Word Error Rate (WER) of the baseline end-to-end speech recognizer, and further improve its performance by combining it with the UDA-based learning method.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2014.11a
• /
• pp.19-20
• /
• 2014

This study is to find out the tensile strength and bonding strength of VAE powder as a preliminary study for the application of the powder to the high strength concrete. The result of the study showed that the compressive strength decreases when more polymers is put into the concrete. On the other hand, it showed that the tensile strength and the bonding strength get improved when the more polymers are put into the concrete. Especially in case of the mixture for high strength concrete, it was found out that more strength is produced than the ordinary concrete.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.8
• /
• pp.23-30
• /
• 2021

In this study, we present a comparative analysis of major autoencoder(AE)-based anomaly detection methods for quality determination in the manufacturing process and a new anomaly discrimination criterion. Due to the characteristics of manufacturing site, anomalous instances are few and their types greatly vary. These properties degrade the performance of an AI-based anomaly detection model using the dataset for both normal and anomalous cases, and incur a lot of time and costs in obtaining additional data for performance improvement. To solve this problem, the studies on AE-based models such as AE and VAE are underway, which perform anomaly detection using only normal data. In this work, based on Convolutional AE, VAE, and Dilated VAE models, statistics on residual images, MSE, and information entropy were selected as outlier discriminant criteria to compare and analyze the performance of each model. In particular, the range value applied to the Convolutional AE model showed the best performance with AUC PRC 0.9570, F1 Score 0.8812 and AUC ROC 0.9548, accuracy 87.60%. This shows a performance improvement of an accuracy about 20%P(Percentage Point) compared to MSE, which was frequently used as a standard for determining outliers, and confirmed that model performance can be improved according to the criteria for determining outliers.

• Journal of Adhesion and Interface
• /
• v.11 no.3
• /
• pp.89-99
• /
• 2010

In this paper, for polymerization of poly(vinyl acetate-co-ethylene) (VAE) by redox system using poly(vinyl alcohol) (PVOH) as emulsifier on the properties of the final emulsion, and pH changes affect the physical properties of the final emulsion was investigated. The results of the molecular weight of PVOH had a dramatic impact on the emulsion properties. The used a low molecular weight of PVOH products was obtained low viscosity and using the high molecular weight of PVOH were obtained high viscosity product. However, changing the pH of the final polymerized product properties for the PVOH obtained different results. Generally, a poly(vinyl acetate) emulsion by a high degree of polymerization and high molecular weight of PVOH was obtained high viscosity of the final emulsion. But, in VAE was lower emulsion viscosity in high pH. This is the molecular weight of the emulsion during the synthesis of PVOH is considered to be affected by degradation. The final viscosity was decreased by grafting ratio and molecular weight were decreased with increasing of pH.

• Journal of Korean Society for Quality Management
• /
• v.49 no.4
• /
• pp.581-594
• /
• 2021

Purpose: The purpose of this study is to compare machine learning models for anomaly detection of mechanical facility equipment and suggest an anomaly detection system for mechanical facility equipment in subway stations. It helps to predict failures and plan the maintenance of facility. Ultimately it aims to improve the quality of facility equipment. Methods: The data collected from Daejeon Metropolitan Rapid Transit Corporation was used in this experiment. The experiment was performed using Python, Scikit-learn, tensorflow 2.0 for preprocessing and machine learning. Also it was conducted in two failure states of the equipment. We compared and analyzed five unsupervised machine learning models focused on model Long Short-Term Memory Variational Autoencoder(LSTM-VAE). Results: In both experiments, change in vibration and current data was observed when there is a defect. When the rotating body failure was happened, the magnitude of vibration has increased but current has decreased. In situation of axis alignment failure, both of vibration and current have increased. In addition, model LSTM-VAE showed superior accuracy than the other four base-line models. Conclusion: According to the results, model LSTM-VAE showed outstanding performance with more than 97% of accuracy in the experiments. Thus, the quality of mechanical facility equipment will be improved if the proposed anomaly detection system is established with this model used.

• Korean Institute of Interior Design Journal
• /
• v.20 no.3
• /
• pp.198-205
• /
• 2011

The study on how visitors of a museum view exhibition is necessary for providing quality experience to the visitors. Previous studies on the movement of visitors of a museum focused on qualitative analysis after the follow-up survey. Therefore, the purpose of this study is to find out various ways to use quantitative analysis methods on the movement of visitors in the museum. Quantitative analysis of the exhibition place and movement of visitors was conducted using programs to produce quantitative results from the space analysis including VAE, VGA, V-ERAM and ESA. VAE and VGA helped to understand the spatial structure and ESA was helpful to predict how the flow of human traffic would be in the museum. If the programs are used all together, it would be easier to quantitatively predict how the How of human traffic would be in an exhibition room. However, this study is in its infancy, so following studies are necessary based on more data and results of analysis in the future.