• 한국컴퓨터정보학회논문지
• /
• 제24권8호
• /
• pp.87-93
• /
• 2019

These days, as interest in the recommendation system with deep learning is increasing, a number of related studies to develop a performance for collaborative filtering through autoencoder, a state-of-the-art deep learning neural network architecture has advanced considerably. The purpose of this study is to propose autoencoder which is used by the recommendation system to predict ratings, and we added more hidden layers to the original architecture of autoencoder so that we implemented deep autoencoder with 3 to 5 hidden layers for much deeper architecture. In this paper, therefore we make a comparison between the performance of them. In this research, we use 2-dimensional arrays and 3-dimensional tensor as the input dataset. As a result, we found a correlation between matrix entry of the 3-dimensional dataset such as item-time and user-time and also figured out that deep autoencoder with extra hidden layers generalized even better performance than autoencoder.

• International Journal of Computer Science & Network Security
• /
• 제21권7호
• /
• pp.159-168
• /
• 2021

Detecting cyber-attacks using machine learning or deep learning is being studied and applied widely in network intrusion detection systems. We noticed that the application of deep learning algorithms yielded many good results. However, because each deep learning model has different architecture and characteristics with certain advantages and disadvantages, so those deep learning models are only suitable for specific datasets or features. In this paper, in order to optimize the process of detecting cyber-attacks, we propose the idea of building a new deep learning network model based on the association and combination of individual deep learning models. In particular, based on the architecture of 2 deep learning models: Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM), we combine them into a combined deep learning network for detecting cyber-attacks based on network traffic. The experimental results in Section IV.D have demonstrated that our proposal using the CNN-LSTM deep learning model for detecting cyber-attacks based on network traffic is completely correct because the results of this model are much better than some individual deep learning models on all measures.

• Geomechanics and Engineering
• /
• 제30권3호
• /
• pp.233-246
• /
• 2022

In this study, the city of Osijek is used as a case study area for low to medium seismicity regions with deep soil over deep geological deposits to determine horizontal PGA values. For this reason, we propose new regional attenuation equations for PGA that can simultaneously capture the effects of deep geology and local soil conditions. A micro-zoning map for the city of Osijek is constructed using the derived empirical scaling equations and compared to all prior seismic hazard estimates for the same area. The findings suggest that the deep soil atop deep geological sediments results in PGA values that are only 6 percent larger than those reported at rock soil sites atop geological rocks. Given the rarity of ground motion records for deep soils atop deep geological layers around the world, we believe this case study is a start toward defining more reliable PGA estimates for similar areas.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권3호
• /
• pp.911-931
• /
• 2021

Recently, most cloud services use Docker container environment to provide their services. However, there are no researches to evaluate the performance of communication libraries for multi-GPU based distributed deep learning in a Docker container environment. In this paper, we propose an efficient communication architecture for multi-GPU based deep learning in a Docker container environment by evaluating the performances of various communication libraries. We compare the performances of the parameter server architecture and the All-reduce architecture, which are typical distributed deep learning architectures. Further, we analyze the performances of two separate multi-GPU resource allocation policies - allocating a single GPU to each Docker container and allocating multiple GPUs to each Docker container. We also experiment with the scalability of collective communication by increasing the number of GPUs from one to four. Through experiments, we compare OpenMPI and MPICH, which are representative open source MPI libraries, and NCCL, which is NVIDIA's collective communication library for the multi-GPU setting. In the parameter server architecture, we show that using CUDA-aware OpenMPI with multi-GPU per Docker container environment reduces communication latency by up to 75%. Also, we show that using NCCL in All-reduce architecture reduces communication latency by up to 93% compared to other libraries.

• Journal of Semiconductor Engineering
• /
• 제1권1호
• /
• pp.31-37
• /
• 2020

We present an approximate DRAM architecture for energy-efficient deep learning. Our key premise is that by bounding memory errors to non-critical information, we can significantly reduce DRAM refresh energy without compromising recognition accuracy of deep neural networks. To validate the key premise, we make extensive Monte-Carlo simulations for several well-known convolutional neural networks such as LeNet, ConvNet and AlexNet with the input of MINIST, CIFAR-10, and ImageNet, respectively. We assume that the highest-order 8-bits (in single precision) and 4-bits (in half precision) are protected from retention errors under the proposed architecture and then, randomly inject bit-errors to unprotected bits with various bit-error-rates. Here, recognition accuracies of the above convolutional neural networks are successfully maintained up to the 10^-5-order bit-error-rate. We simulate DRAM energy during inference of the above convolutional neural networks, where the proposed architecture shows the possibility of considerable energy saving up to 10 ~ 37.5% of total DRAM energy.

• 한국산업융합학회 논문집
• /
• 제25권4_2호
• /
• pp.687-697
• /
• 2022

An automated system is needed for the effectiveness of non-destructive testing. In order to utilize the radiographic testing data accumulated in the film, the types of welding defects were classified into 9 and the shape of defects were analyzed. Data was preprocessed to use deep learning with high performance in image classification, and a combination of one-stage/two-stage method and convolutional neural networks/Transformer backbone was compared to confirm a model suitable for welding defect detection. The combination of two-stage, which can learn step-by-step, and deep-layered CNN backbone, showed the best performance with mean average precision 0.868.

• ETRI Journal
• /
• 제42권4호
• /
• pp.505-517
• /
• 2020

The increasing size and complexity of deep neural networks (DNNs) necessitate the development of efficient high-performance accelerators. An efficient memory structure and operating scheme provide an intuitive solution for high-performance accelerators along with dataflow control. Furthermore, the processing of various neural networks (NNs) requires a flexible memory architecture, programmable control scheme, and automated optimizations. We first propose an efficient architecture with flexibility while operating at a high frequency despite the large memory and PE-array sizes. We then improve the efficiency and usability of our architecture by automating the optimization algorithm. The experimental results show that the architecture increases the data reuse; a diagonal write path improves the performance by 1.44× on average across a wide range of NNs. The automated optimizations significantly enhance the performance from 3.8× to 14.79× and further provide usability. Therefore, automating the optimization as well as designing an efficient architecture is critical to realizing high-performance DNN accelerators.

• 전기전자학회논문지
• /
• 제25권4호
• /
• pp.644-649
• /
• 2021

Self-Imitation Learning은 간단한 비활성 정책 actor-critic 알고리즘으로써 에이전트가 과거의 좋은 경험을 활용하여 최적의 정책을 찾을 수 있도록 해준다. 그리고 actor-critic 구조를 갖는 강화학습 알고리즘에 결합되어 다양한 환경들에서 알고리즘의 상당한 개선을 보여주었다. 하지만 Self-Imitation Learning이 강화학습에 큰 도움을 준다고 하더라도 그 적용 분야는 actor-critic architecture를 가지는 강화학습 알고리즘으로 제한되어 있다. 본 논문에서 Self-Imitation Learning의 알고리즘을 가치 기반 강화학습 알고리즘인 DQN에 적용하는 방법을 제안하고, Self-Imitation Learning이 적용된 DQN 알고리즘의 학습을 다양한 환경에서 진행한다. 아울러 그 결과를 기존의 결과와 비교함으로써 Self-Imitation Leaning이 DQN에도 적용될 수 있으며 DQN의 성능을 개선할 수 있음을 보인다.

• 대한건축학회논문집:계획계
• /
• 제36권1호
• /
• pp.61-68
• /
• 2020

The aim of the paper is to verify the hypothesis that 'Deep Contextualism can be applied to architecture,' developed through Stephan Pepper's Contextualism. This hypothesis is verified by Peter Zumthor's architectures. 'Deep Contextualism' has 'the third context' between human and natural context. The third so-called "ready-to-hand" is reactivated according to contingency and creativity in Being, 'Being' is continuously revitalized. It means that quality in 'the third context, or a sense of ready-to-hand is creatively and unceasingly transformed into a new texture by creativity and contingency in Being. Also Zumthor and Heidegger agrees to a sense of ready-to-hand. Through creativity and contingency in Being his architecture widens and deepens. Zumthor's atmosphere is world 4, and his architecture is world 3 which is an embodiment of world 4 or his air.

• Ocean Systems Engineering
• /
• 제10권4호
• /
• pp.415-434
• /
• 2020

Detecting objects is important for the safe operation of ships, and enables collision avoidance, risk detection, and autonomous sailing. This study proposes a ship detection method from images and videos taken at sea using one of the state-of-the-art deep neural network-based object detection algorithms. A deep learning model is trained using a public maritime dataset, and results show it can detect all types of floating objects and classify them into ten specific classes that include a ship, speedboat, and buoy. The proposed deep learning model is compared to a universal trained model that detects and classifies objects into general classes, such as a person, dog, car, and boat, and results show that the proposed model outperforms the other in the detection of maritime objects. Different deep neural network structures are then compared to obtain the best detection performance. The proposed model also shows a real-time detection speed of approximately 30 frames per second. Hence, it is expected that the proposed model can be used to detect maritime objects and reduce risks while at sea.