• The Journal of Korea Robotics Society
• /
• v.17 no.2
• /
• pp.133-141
• /
• 2022

This paper introduces model compression algorithms which make a deep neural network smaller and faster for embedded systems. The model compression algorithms can be largely categorized into pruning, quantization and knowledge distillation. In this study, gradual pruning, quantization aware training, and knowledge distillation which learns the activation boundary in the hidden layer of the teacher neural network are integrated. As a large deep neural network is compressed and accelerated by these algorithms, embedded computing boards can run the deep neural network much faster with less memory usage while preserving the reasonable accuracy. To evaluate the performance of the compressed neural networks, we evaluate the size, latency and accuracy of the deep neural network, DenseNet201, for image classification with CIFAR-10 dataset on the NVIDIA Jetson Xavier.

• Journal of information and communication convergence engineering
• /
• v.21 no.1
• /
• pp.9-16
• /
• 2023

The natural language on social network platforms has a certain front-to-back dependency in structure, and the direct conversion of Chinese text into a vector makes the dimensionality very high, thereby resulting in the low accuracy of existing text classification methods. To this end, this study establishes a deep learning model that combines a big data ultra-deep convolutional neural network (UDCNN) and long short-term memory network (LSTM). The deep structure of UDCNN is used to extract the features of text vector classification. The LSTM stores historical information to extract the context dependency of long texts, and word embedding is introduced to convert the text into low-dimensional vectors. Experiments are conducted on the social network platforms Sogou corpus and the University HowNet Chinese corpus. The research results show that compared with CNN + rand, LSTM, and other models, the neural network deep learning hybrid model can effectively improve the accuracy of text classification.

• Journal of information and communication convergence engineering
• /
• v.21 no.2
• /
• 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.

• Communications for Statistical Applications and Methods
• /
• v.29 no.2
• /
• pp.161-176
• /
• 2022

There has been active research in image classification using deep learning convolutional neural network (CNN) models. ImageNet large-scale visual recognition challenge (ILSVRC) (2010-2017) was one of the most important competitions that boosted the development of efficient deep learning algorithms. This paper introduces and compares six monumental models that achieved high prediction accuracy in ILSVRC. First, we provide a review of the models to illustrate their unique structure and characteristics of the models. We then compare those models under a unified framework. For this reason, additional devices that are not crucial to the structure are excluded. Four popular data sets with different characteristics are then considered to measure the prediction accuracy. By investigating the characteristics of the data sets and the models being compared, we provide some insight into the architectural features of the models.

• International journal of advanced smart convergence
• /
• v.12 no.2
• /
• pp.117-126
• /
• 2023

To improve the overall performance and realism of your game, it is important to calculate the trajectory of a projectile accurately and quickly. One way to increase realism is to use a ballistic model that takes into account factors such as air resistance, density, and wind when calculating a projectile's trajectory. However, the more these factors are taken into account, the more computationally time-consuming and expensive it becomes, creating a trade-off between overall performance and efficiency. Therefore, we present an optimal solution to find a balance between ballistic model accuracy and computation time. We perform ballistic calculations using numerical methods such as Euler, Velocity Verlet, RK2, RK4, and Akima interpolation, and measure and compare the computation time, memory usage (RSS, Resident Set Size), and accuracy of each method. We show developers how to implement more accurate and efficient ballistic models and help them choose the right computational method for their numerical applications.

• Journal of Information Processing Systems
• /
• v.19 no.5
• /
• pp.688-701
• /
• 2023

The conventional methods of network intrusion detection system (NIDS) cannot measure the trend of intrusiondetection targets effectively, which lead to low detection accuracy. In this study, a NIDS method which based on a deep neural network in a big-data environment is proposed. Firstly, the entire framework of the NIDS model is constructed in two stages. Feature reduction and anomaly probability output are used at the core of the two stages. Subsequently, a convolutional neural network, which encompasses a down sampling layer and a characteristic extractor consist of a convolution layer, the correlation of inputs is realized by introducing bidirectional long short-term memory. Finally, after the convolution layer, a pooling layer is added to sample the required features according to different sampling rules, which promotes the overall performance of the NIDS model. The proposed NIDS method and three other methods are compared, and it is broken down under the conditions of the two databases through simulation experiments. The results demonstrate that the proposed model is superior to the other three methods of NIDS in two databases, in terms of precision, accuracy, F1- score, and recall, which are 91.64%, 93.35%, 92.25%, and 91.87%, respectively. The proposed algorithm is significant for improving the accuracy of NIDS.

• International Journal of Advanced Culture Technology
• /
• v.11 no.4
• /
• pp.393-405
• /
• 2023

Accurate prediction of wind speed and power is vital for enhancing the efficiency of wind energy systems. Numerous solutions have been implemented to date, demonstrating their potential to improve forecasting. Among these, deep learning is perceived as a revolutionary approach in the field. However, despite their effectiveness, the noise present in the collected data remains a significant challenge. This noise has the potential to diminish the performance of these algorithms, leading to inaccurate predictions. In response to this, this study explores a novel feature engineering approach. This approach involves altering the data input shape in both Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) and Autoregressive models for various forecasting horizons. The results reveal substantial enhancements in model resilience against noise resulting from step increases in data. The approach could achieve an impressive 83% accuracy in predicting unseen data up to the 24th steps. Furthermore, this method consistently provides high accuracy for short, mid, and long-term forecasts, outperforming the performance of individual models. These findings pave the way for further research on noise reduction strategies at different forecasting horizons through shape-wise feature engineering.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.20 no.1
• /
• pp.11-18
• /
• 2020

This study aims to complement the poor performance of existing context-aware authentication techniques in the mobile environment. The data used are GPS, Call Detail Record(CDR) and app usage. locational classification according to GPS density was implemented in order to distinguish other people in populated areas in the processing of GPS. It also handles missing values that may occur in data collection. The authentication model consists of two long-short term memory(LSTM) and one Artificial Neural Network(ANN) that aggregates the results, which produces authentication scores. In this paper, we compare the accuracy of this technique with that of other studies. Then compare the number of authentication attempts required to detect someone else's authentication. As a result, we achieved an average 11.6% improvement in accuracy and faster detection of approximately 60% of the experimental data.

• Journal of Power Electronics
• /
• v.11 no.2
• /
• pp.228-236
• /
• 2011

The ${\Delta}\;V_{10}$ or 10-Hz flicker index, as a common method of measurement of voltage flicker severity in power systems, requires a high computational cost and a large amount of memory. In this paper, for measuring the ${\Delta}\;V_{10}$ index, a new method based on the Adaline (adaptive linear neuron) system, the FFT (fast Fourier transform), and the PSO (particle swarm optimization) algorithm is proposed. In this method, for reducing the sampling frequency, calculations are carried out on the envelope of a power system voltage that contains a flicker component. Extracting the envelope of the voltage is implemented by the Adaline system. In addition, in order to increase the accuracy in computing the flicker components, the PSO algorithm is used for reducing the spectral leakage error in the FFT calculations. Therefore, the proposed method has a lower computational cost in FFT computation due to the use of a smaller sampling window. It also requires less memory since it uses the envelope of the power system voltage. Moreover, it shows more accuracy because the PSO algorithm is used in the determination of the flicker frequency and the corresponding amplitude. The sensitivity of the proposed method with respect to the main frequency drift is very low. The proposed algorithm is evaluated by simulations. The validity of the simulations is proven by the implementation of the algorithm with an ARM microcontroller-based digital system. Finally, its function is evaluated with real-time measurements.

• Tunnel and Underground Space
• /
• v.28 no.3
• /
• pp.277-291
• /
• 2018

LSTM (Long Short-Term Memory) algorithm which is a kind of recurrent neural network was used to establish a model to predict the stress-strain curve of an material under uniaxial compression. The model was established from the stress-strain data from uniaxial compression tests of silica-gypsum specimens. After training the model, it can predict the behavior of the material up to the failure state by using an early stage of stress-strain curve whose stress is very low. Because the LSTM neural network predict a value by using the previous state of data and proceed forward step by step, a higher error was found at the prediction of higher stress state due to the accumulation of error. However, this model generally predict the stress-strain curve with high accuracy. The accuracy of both LSTM and tangential prediction models increased with increased length of input data, while a difference in performance between them decreased as the amount of input data increased. LSTM model showed relatively superior performance to the tangential prediction when only few input data was given, which enhanced the necessity for application of the model.