• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.1-17
• /
• 2017

The deep learning framework is software designed to help develop deep learning models. Some of its important functions include "automatic differentiation" and "utilization of GPU". The list of popular deep learning framework includes Caffe (BVLC) and Theano (University of Montreal). And recently, Microsoft's deep learning framework, Microsoft Cognitive Toolkit, was released as open-source license, following Google's Tensorflow a year earlier. The early deep learning frameworks have been developed mainly for research at universities. Beginning with the inception of Tensorflow, however, it seems that companies such as Microsoft and Facebook have started to join the competition of framework development. Given the trend, Google and other companies are expected to continue investing in the deep learning framework to bring forward the initiative in the artificial intelligence business. From this point of view, we think it is a good time to compare some of deep learning frameworks. So we compare three deep learning frameworks which can be used as a Python library. Those are Google's Tensorflow, Microsoft's CNTK, and Theano which is sort of a predecessor of the preceding two. The most common and important function of deep learning frameworks is the ability to perform automatic differentiation. Basically all the mathematical expressions of deep learning models can be represented as computational graphs, which consist of nodes and edges. Partial derivatives on each edge of a computational graph can then be obtained. With the partial derivatives, we can let software compute differentiation of any node with respect to any variable by utilizing chain rule of Calculus. First of all, the convenience of coding is in the order of CNTK, Tensorflow, and Theano. The criterion is simply based on the lengths of the codes and the learning curve and the ease of coding are not the main concern. According to the criteria, Theano was the most difficult to implement with, and CNTK and Tensorflow were somewhat easier. With Tensorflow, we need to define weight variables and biases explicitly. The reason that CNTK and Tensorflow are easier to implement with is that those frameworks provide us with more abstraction than Theano. We, however, need to mention that low-level coding is not always bad. It gives us flexibility of coding. With the low-level coding such as in Theano, we can implement and test any new deep learning models or any new search methods that we can think of. The assessment of the execution speed of each framework is that there is not meaningful difference. According to the experiment, execution speeds of Theano and Tensorflow are very similar, although the experiment was limited to a CNN model. In the case of CNTK, the experimental environment was not maintained as the same. The code written in CNTK has to be run in PC environment without GPU where codes execute as much as 50 times slower than with GPU. But we concluded that the difference of execution speed was within the range of variation caused by the different hardware setup. In this study, we compared three types of deep learning framework: Theano, Tensorflow, and CNTK. According to Wikipedia, there are 12 available deep learning frameworks. And 15 different attributes differentiate each framework. Some of the important attributes would include interface language (Python, C ++, Java, etc.) and the availability of libraries on various deep learning models such as CNN, RNN, DBN, and etc. And if a user implements a large scale deep learning model, it will also be important to support multiple GPU or multiple servers. Also, if you are learning the deep learning model, it would also be important if there are enough examples and references.

• International journal of advanced smart convergence
• /
• v.2 no.2
• /
• pp.7-9
• /
• 2013

Current physics-based simulation is an important tool in the fluid animation. However some problems require a new change to current research trends which depend only on the simulation. The ultimate goal of this project is to obtain information of flow example, analyze an example through machine learning and the novel fluid animation reconfigure without physical simulation.

• Proceedings of the IEEK Conference
• /
• 2000.06a
• /
• pp.115-118
• /
• 2000

In order to control the flow of traffics in ATM networks and optimize the usage of network resources, an efficient control mechanism is necessary to cope with congestion and prevent the degradation of network performance caused by congestion. To effectively control traffic in UNI(User Network Interface) stage, we proposed algorithm of integrated model using on-line teaming neural network for CAC(Call Admission Control) and UPC(Usage Parameter Control). Simulation results will show that the proposed adaptive algorithm uses of network resources efficiently and satisfies QoS for the various kinds of traffics.

• Journal of Information Processing Systems
• /
• v.6 no.3
• /
• pp.307-322
• /
• 2010

Traffic classification is an essential task for network management. Many researchers have paid attention to initial sub-flow features based classifiers for traffic classification. However, the existing classifiers cannot classify traffic effectively in mobile IP networks. The classifiers depend on initial sub-flows, but they cannot always capture the sub-flows at a point of attachment for a variety of elements because of seamless mobility. Thus the ideal classifier should be capable of traffic classification based on not only initial sub-flows but also various types of sub-flows. In this paper, we propose a classifiable sub-flow selection method to realize the ideal classifier. The experimental results are so far promising for this research direction, even though they are derived from a reduced set of general applications and under relatively simplifying assumptions. Altogether, the significant contribution is indicating the feasibility of the ideal classifier by selecting not only initial sub-flows but also transition sub-flows.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2006.11a
• /
• pp.289-292
• /
• 2006

In this paper, the hybrid algorithm of Support Vector Machine md Artificial Neural Network is used for the defect diagnostics algorithm for the aircraft turbo-shaft engine. The results of learning of ANN, especially, accuracy or speed of convergence are sensitive to the number of data, so a comparison between design point and off-design area, especially, Mach number and fuel flow variable area, is essential research. From application results for diagnostics of gas turbine engine, it was confirmed that the hybrid algorithm could detect well in the of-design area as well as design point.

• The Journal of Pediatrics of Korean Medicine
• /
• v.23 no.3
• /
• pp.241-262
• /
• 2009

Objectives This study is designed to investigate the effects of Jujadokseo-hwan on the brain ability and inducing oxidative stresses. Methods We measured the changes of regional cerebral blood flow and mean arterial blood pressure. Then we analyzed histological examination, immunohistochemistric response and anti-oxidant activity of Jujadokseo-hwan. Results 1. Treatment of Jujadokseo-hwan significantly increased regional cerebral blood flow but decreased mean arterial blood pressure. 2. Treatment of Jujadokseo-hwan-induced increase of regional cerebral blood flow was significantly inhibited by pretreatment with indomethacin (1 mg/kg, i.p.), an inhibitor of cyclooxygenase. 3. In histological examination through TTC stain, group I was no change, but group II showed that discolored in the most cortical part. Group III showed that decreased discolor in the cortical part. 4. In immunohistochemistric response of BDNF, group II showed that lower response effect. Group III showed that increase response effect. 5. Treatment of Jujadokseo-hwan increased proliferation rates of Glial cell effectively 6. Treatment of Jujadokseo-hwan accelerated proliferation rates of C6 cells in vitro. In addition, protective effects on cell death induced by paraquat, rotenone and hydrogen peroxide. In addition, activity of SOD were increased by treatment with Jujadokseo-hwan. Conclusions In conclusion, Jujadokseo-hwan can improve of the brain ability, learning ability, memory ability and induce ischemic brain injuries.

• International Journal of Contents
• /
• v.15 no.4
• /
• pp.8-15
• /
• 2019

This paper presents a solution of the 'Quick, Draw! Doodle Recognition Challenge' hosted by Google. Doodles are drawings comprised of concrete representational meaning or abstract lines creatively expressed by individuals. In this challenge, a doodle is presented as a sequence of sketches. From the view of at the sketch level, to learn the pattern of strokes representing a doodle, we propose a sequential model stacked with multiple convolution layers and Long Short-Term Memory (LSTM) cells following the attention mechanism [15]. From the view at the image level, we use multiple models pre-trained on ImageNet to recognize the doodle. Finally, an ensemble and a post-processing method using the minimum cost flow algorithm are introduced to combine multiple models in achieving better results. In this challenge, our solutions garnered 11th place among 1,316 teams. Our performance was 0.95037 MAP@3, only 0.4% lower than the winner. It demonstrates that our method is very competitive. The source code for this competition is published at: https://github.com/ngxbac/Kaggle-QuickDraw.

• Journal of the Korean Society of Visualization
• /
• v.22 no.1
• /
• pp.49-60
• /
• 2024

The purpose of this study establishes the deep neural network (DNN) and Decoder models to predict the flow and thermal fields of three-dimensional wavy wings as a passive flow control. The wide ranges of the wavy geometric parameters of wave amplitude and wave number are considered for the various the angles of attack and the aspect ratios of a wing. The huge dataset for training and test of the deep learning models are generated using computational fluid dynamics (CFD). The DNN and Decoder models exhibit quantitatively accurate predictions for aerodynamic coefficients and Nusselt numbers, also qualitative pressure, limiting streamlines, and Nusselt number distributions on the surface. Particularly, Decoder model regenerates the important flow features of tiny vortices in the valleys, which makes a delay of the stall. Also, the spiral vortical formation is realized by the Decoder model, which enhances the lift.

• Journal of Korea Water Resources Association
• /
• v.51 no.12
• /
• pp.1207-1216
• /
• 2018

Discharge or water level predictions at tidally affected river reaches are currently still a great challenge in hydrological practices. This research aims to predict water level of the tide dominated site, Jamsu bridge in the Han River downstream. Physics-based hydrodynamic approaches are sometimes not applicable for water level prediction in such a tidal river due to uncertainty sources like rainfall forecasting data. In this study, TensorFlow deep learning framework was used to build a deep neural network based LSTM model and its applications. The LSTM model was trained based on 3 data sets having 10-min temporal resolution: Paldang dam release, Jamsu bridge water level, predicted tidal level for 6 years (2011~2016) and then predict the water level time series given the six lead times: 1, 3, 6, 9, 12, 24 hours. The optimal hyper-parameters of LSTM model were set up as follows: 6 hidden layers number, 0.01 learning rate, 3000 iterations. In addition, we changed the key parameter of LSTM model, sequence length, ranging from 1 to 6 hours to test its affect to prediction results. The LSTM model with the 1 hr sequence length led to the best performing prediction results for the all cases. In particular, it resulted in very accurate prediction: RMSE (0.065 cm) and NSE (0.99) for the 1 hr lead time prediction case. However, as the lead time became longer, the RMSE increased from 0.08 m (1 hr lead time) to 0.28 m (24 hrs lead time) and the NSE decreased from 0.99 (1 hr lead time) to 0.74 (24 hrs lead time), respectively.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.9
• /
• pp.49-57
• /
• 2022

Currently, the problem of food shortage is emerging in our society due to climate problems and an increase population in the world. As a solution to this problem, we propose a multi-remote control smart farm that combines artificial intelligence (AI) and information and communication technology (ICT) technologies. The proposed smart farm integrates ICT technology to remotely control and manage crops without restrictions in space and time, and to multi-control the growing environment of crops. In addition, using Arduino and deep-learning technology, a smart farm capable of multiple control through a smart-phone application (APP) was proposed, and Ai technology with various data securing and diagnosis functions while observing crop growth in real-time was included. Various sensors in the smart farm are controlled by using the Arduino, and the data values of the sensors are stored in the built database, so that the user can check the stored data with the APP. For multiple control for multiple crops, each LED, COOLING FAN, and WATER PUMP for two or more growing environments were applied so that the user could control it conveniently. And by implementing an APP that diagnoses the growth stage through the Tensor-Flow framework using deep-learning technology, we developed an application that helps users to easily diagnose the growth status of the current crop.