• Steel and Composite Structures
• /
• v.47 no.6
• /
• pp.759-779
• /
• 2023

The paper proposes two hybrid metaheuristic optimization and artificial neural network (ANN) methods for the close prediction of the ultimate axial compressive capacity of concentrically loaded concrete filled double skin steel tube (CFDST) columns. Two metaheuristic optimization, namely genetic algorithm (GA) and particle swarm optimization (PSO), approaches enable the dynamic training architecture underlying an ANN model by optimizing the number and sizes of hidden layers as well as the weights and biases of the neurons, simultaneously. The former is termed as GA-ANN, and the latter as PSO-ANN. These techniques utilize the gradient-based optimization with Bayesian regularization that enhances the optimization process. The proposed GA-ANN and PSO-ANN methods construct the predictive ANNs from 125 available experimental datasets and present the superior performance over standard ANNs. Both the hybrid GA-ANN and PSO-ANN methods are encoded within a user-friendly graphical interface that can reliably map out the accurate ultimate axial compressive capacity of CFDST columns with various geometry and material parameters.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1191-1194
• /
• 2022

인터넷의 발전으로 수많은 이미지와 비디오를 손쉽게 이용할 수 있게 되었다. 이미지와 비디오 데이터의 양이 기하급수적으로 증가함에 따라, JPEG, HEVC, VVC 등 이미지와 비디오를 효율적으로 저장하기 위한 부호화 기술들이 등장했다. 최근에는 인공신경망을 활용한 학습 기반 모델이 발전함에 따라, 이를 활용한 이미지 및 비디오 압축 기술에 관한 연구가 빠르게 진행되고 있다. NNIC (Neural Network based Image Coding)는 이러한 학습 가능한 인공신경망 기반 이미지 부호화 기술을 의미한다. 본 논문에서는 NNIC 모델과 인공신경망 기반의 초해상화(Super Resolution) 모델을 합동훈련하여 기존 NNIC 모델보다 더 높은 성능을 보일 수 있는 방법을 제시한다. 먼저 NNIC 인코더(Encoder)에 이미지를 입력하기 전 다운 스케일링(Down Scaling)으로 쌍삼차보간법을 사용하여 이미지의 화소를 줄인 후 부호화(Encoding)한다. NNIC 디코더(Decoder)를 통해 부호화된 이미지를 복호화(Decoding)하고 업 스케일링으로 초해상화를 통해 복호화된 이미지를 원본 이미지로 복원한다. 이때 NNIC 모델과 초해상화 모델을 합동훈련한다. 결과적으로 낮은 비트량에서 더 높은 성능을 볼 수 있는 가능성을 보았다. 또한 합동훈련을 함으로써 전체 성능의 향상을 보아 학습 시간을 늘리고, 압축 잡음을 위한 초해상화 모델을 사용한다면 기존의 NNIC 보다 나은 성능을 보일 수 있는 가능성을 시사한다.

• Steel and Composite Structures
• /
• v.45 no.2
• /
• pp.281-291
• /
• 2022

Surmounting complexities in analyzing the mechanical parameters of concrete entails selecting an appropriate methodology. This study integrates a novel metaheuristic technique, namely satin bowerbird optimizer (SBO) with artificial neural network (ANN) for predicting uniaxial compressive strength (UCS) of concrete. For this purpose, the created hybrid is trained and tested using a relatively large dataset collected from the published literature. Three other new algorithms, namely Henry gas solubility optimization (HGSO), sunflower optimization (SFO), and vortex search algorithm (VSA) are also used as benchmarks. After attaining a proper population size for all algorithms, the Utilizing various accuracy indicators, it was shown that the proposed ANN-SBO not only can excellently analyze the UCS behavior, but also outperforms all three benchmark hybrids (i.e., ANN-HGSO, ANN-SFO, and ANN-VSA). In the prediction phase, the correlation indices of 0.87394, 0.87936, 0.95329, and 0.95663, as well as mean absolute percentage errors of 15.9719, 15.3845, 9.4970, and 8.0629%, calculated for the ANN-HGSO, ANN-SFO, ANN-VSA, and ANN-SBO, respectively, manifested the best prediction performance for the proposed model. Also, the ANN-VSA achieved reliable results as well. In short, the ANN-SBO can be used by engineers as an efficient non-destructive method for predicting the UCS of concrete.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.3
• /
• pp.89-100
• /
• 2023

Recently, on-device artificial intelligence (AI) solutions using mobile devices and embedded edge devices have emerged in various fields, such as computer vision, to address network traffic burdens, low-energy operations, and security problems. Although vision transformer deep learning models have outperformed conventional convolutional neural network (CNN) models in computer vision, they require more computations and parameters than CNN models. Thus, they are not directly applicable to embedded edge devices with limited hardware resources. Many researchers have proposed various model compression methods or lightweight architectures for vision transformers; however, there are only a few studies evaluating the effects of model compression techniques of vision transformers on performance. Regarding this problem, this paper presents a performance evaluation of vision transformers on embedded platforms. We investigated the behaviors of three vision transformers: DeiT, LeViT, and MobileViT. Each model performance was evaluated by accuracy and inference time on edge devices using the ImageNet dataset. We assessed the effects of the quantization method applied to the models on latency enhancement and accuracy degradation by profiling the proportion of response time occupied by major operations. In addition, we evaluated the performance of each model on GPU and EdgeTPU-based edge devices. In our experimental results, LeViT showed the best performance in CPU-based edge devices, and DeiT-small showed the highest performance improvement in GPU-based edge devices. In addition, only MobileViT models showed performance improvement on EdgeTPU. Summarizing the analysis results through profiling, the degree of performance improvement of each vision transformer model was highly dependent on the proportion of parts that could be optimized in the target edge device. In summary, to apply vision transformers to on-device AI solutions, either proper operation composition and optimizations specific to target edge devices must be considered.

• International Journal of Automotive Technology
• /
• v.4 no.1
• /
• pp.47-55
• /
• 2003

The Center for Transportation Research at the Argonne National Laboratory (ANL) supports the DOE by evaluating advanced automotive technologies in a systems context. ha has developed a unique set of compatible simulation tools and test equipment to perform an integrated systems analysis project from modeling through hardware testing and validation. This project utilized these capabilities to demonstrate the trade-off in fuel economy and Oxides of Nitrogen (NOx) emissions in a so-called ‘pre-transmission’ parallel hybrid powertrain. The powertrain configuration (in simulation and on the dynamometer) consists of a Compression Ignition Direct Ignition (CIDI) engine, a Continuously Variable Transmission (CVT) and an electric drive motor coupled to the CVT input shaft. The trade-off is studied in a simulated environment using PSAT with different controllers (fuzzy logic and rule based) and engine models (neural network and steady state models developed from ANL data).

• Journal of the Korean Society for Precision Engineering
• /
• v.23 no.1 s.178
• /
• pp.136-144
• /
• 2006

Aluminum engine piston is manufactured by thixoforging according to forming variables. It is very important to find effects of forming variables on final products in thixoferging. In order to find the effects, however, many researchers and industrial technicians have depended upon too many types of experiments. In this study, the process parameters which have influences on thixofurging process of aluminum automotive engine piston are found by a statistical method and the correlation equations between the process parameters and quality of product are approximated through the surface response analysis. Forming variables such as initial solid fraction, die temperature, and compression holding time are considered fur manufacturing aluminum engine piston by thixofurging. Hardness and microstructure are inspected so that optimal forming condition is found by the statistical approach.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.54 no.7
• /
• pp.332-338
• /
• 2005

PD(Partial Discharge) signal emitted from PD sources has their intrinsic features in the region of time and frequency STFT(Short Time Fourier Transform) shows time-frequency distribution at the same time. 2-Dimensional matrices(33$\times$77) from STFT for PD pulse signals are a good feature vectors and can be decreased in dimension by wavelet 2D data compression technique. Decreased feature vectors(13$\times$24) were used as inputs of Back-propagation ANN(Artificial Neural Network) for discrimination of Multi-PD sources(air discharge sources(3), surface discharge(1)). They are a good feature vectors for discriminating Multi-PD sources in the air.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2006.05a
• /
• pp.429-433
• /
• 2006

High Temperature deformation behavior of newly developed beta-gamma TiAl alloy was investigated in this study. The optimum processing condition was investigated with the aid of Dynamic Materials Model (DMM). Processing maps representing the efficiency of power dissipation for microstructural evolution and instability were constructed utilizing the results of hot compression test at temperatures ranging from $1000^{\circ}C$ to $1200^{\circ}C$ and strain rate ranging from $10^{-4}/s$ to $10^2/s$. The Artificial Neural Network (ANN) simulation was adopted to consider the deformation heating. With the help of processing map and microstructural analysis, the optimum processing condition was presented and the role of $\beta$ phase was also discussed in this study.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.1
• /
• pp.10-22
• /
• 1993

In this paper, we proposed a Neuro-FAX algorithm having high compression rate and good reconstruction capability in spite of noise and fonts. This algorithm processes the character part and the image part seperately. In the character part, we recognized each characters in document using neural networks, and transmitted the information recognized. And we transmitted the image part as it is by the conventional method. With character set in receiving terminal. it can produce nice document of noise free characters and different font.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.133-135
• /
• 2020

최근 CNN(Convolutional Neural Network)은 영상 분류, 객체 인식 등 다양한 비전 분야에서 우수한 성능을 보여주고 있으나, CNN 모델의 계산량 및 메모리가 매우 커짐에 따라 모바일 또는 IoT(lnternet of Things) 장치와 같은 저전력 환경에 적용되기에는 제한이 따른다. 따라서, CNN 모델의 임무 성능을 유지하연서 네트워크 모델을 압축하는 기법들이 연구되고 있다. 본 논문에서는 행렬 분해 기술인 저계수행렬 근사(Low-rank approximation)와 CP(Canonical Polyadic) 분해 기법을 결합하여 CNN 모델을 압축하는 기법을 제안한다. 제안하는 기법은 계층의 유형에 상관없이 하나의 행렬분해 기법만을 적용하는 기존의 기법과 달리 압축 성능을 높이기 위하여 CNN의 계층 타입에 따라 두 가지 분해 기법을 선택적으로 적용한다. 제안기법의 성능검증을 위하여 영상 분류 CNN 모델인 VGG-16, ResNet50, 그리고 MobileNetV2 모델 압축에 적용하였고, 모델의 계층 유형에 따라 두 가지의 분해 기법을 선택적으로 적용함으로써 저계수행렬 근사 기법만 적용한 경우 보다 1.5~12.1 배의 동일한 압축율에서 분류 성능이 향상됨을 확인하였다.