• Journal of the Korean Society of Systems Engineering
• /
• v.18 no.2
• /
• pp.75-93
• /
• 2022

Machine learning (ML) data-driven meta-model is proposed as a surrogate model to reduce the excessive computational cost of the physics-based model and facilitate the real-time prediction of a nuclear power plant's transient response. To forecast the transient response three machine learning (ML) meta-models based on recurrent neural networks (RNNs); specifically, Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), and a sequence combination of Convolutional Neural Network (CNN) and LSTM are developed. The chosen accident scenario is a control element assembly withdrawal at power concurrent with the Loss Of Offsite Power (LOOP). The transient response was obtained using the best estimate thermal hydraulics code, MARS-KS, and cross-validated against the Design and control document (DCD). DAKOTA software is loosely coupled with MARS-KS code via a python interface to perform the Best Estimate Plus Uncertainty Quantification (BEPU) analysis and generate a time series database of the system response to train, test and validate the ML meta-models. Key uncertain parameters identified as required by the CASU methodology were propagated using the non-parametric Monte-Carlo (MC) random propagation and Latin Hypercube Sampling technique until a statistically significant database (181 samples) as required by Wilk's fifth order is achieved with 95% probability and 95% confidence level. The three ML RNN models were built and optimized with the help of the Talos tool and demonstrated excellent performance in forecasting the most probable NPP transient response. This research was guided by the Systems Engineering (SE) approach for the systematic and efficient planning and execution of the research.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2010.05a
• /
• pp.457-458
• /
• 2010

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.5
• /
• pp.395-401
• /
• 2023

Sound Event Detection (SED) aims to identify not only sound category but also time interval for target sounds in an audio waveform. It is a critical technique in field of acoustic surveillance system and monitoring system. Recently, various models have introduced through Detection and Classification of Acoustic Scenes and Events (DCASE) Task 4. This paper explored how to design optimal parameters of DenseNet based model, which has led to outstanding performance in other recognition system. In experiment, DenseRNN as an SED model consists of DensNet-BC and bi-directional Gated Recurrent Units (GRU). This model is trained with Mean teacher model. With an event-based f-score, evaluation is performed depending on parameters, related to model architecture as well as model training, under the assessment protocol of DCASE task4. Experimental result shows that the performance goes up and has been saturated to near the best. Also, DenseRNN would be trained more effectively without dropout technique.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2000.05a
• /
• pp.322-328
• /
• 2000

This study discusses cognition as a computable mapping in cognitive system and relates G del's Incompleteness Theorems to the computability of cognition from a metamathematical perspective. Understanding cognition as a from of computation requires not only Turing machine models but also neural network models. In previous studies of computation by cognitive systems, it is remarkable to note how little serious attention has been given to the issue of computation by neural networks with respect to G del's Incompleteness Theorems. To address this problem, first, we introduce a definition of cognition and cognitive science. Second, we deal with G del's view of computability, incompleteness and speed-up theorems, and then we interpret G del's disjunction on the mind and the machine. Third, we discuss cognition as a Turing computable function and its relation to G del's incompleteness. Finally, we investigate cognition as a neural computable function and its relation to G del's incompleteness. The results show that a second-order representing system can be implemented by a finite recurrent neural network. Hence one cannot prove the consistency of such neural networks in terms of first-order theories. Neural computability, theoretically, is beyond the computational incompleteness of Turing machines. If cognition is a neural computable function, then G del's incompleteness result does not limit the compytational capability of cognition in humans or in artifacts.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.4
• /
• pp.276-287
• /
• 2014

Neural fuzzy system (NFS) is basically a fuzzy system that has been equipped with learning capability adapted from the learning idea used in neural networks. Due to their outstanding system modeling capability, NFS have been widely employed in various applications. In this article, we intend to discuss several ideas regarding the learning of NFS for modeling systems. The first issue discussed here is about structure learning techniques. Various ideas used in the literature are introduced and discussed. The second issue is about the use of recurrent networks in NFS to model dynamic systems. The discussion about the performance of such systems will be given. It can be found that such a delay feedback can only bring one order to the system not all possible order as claimed in the literature. Finally, the mechanisms and relative learning performance of with the use of the recursive least squares (RLS) algorithm are reported and discussed. The analyses will be on the effects of interactions among rules. Two kinds of systems are considered. They are the strict rules and generalized rules and have difference variances for membership functions. With those observations in our study, several suggestions regarding the use of the RLS algorithm in NFS are presented.

• Asia pacific journal of information systems
• /
• v.29 no.4
• /
• pp.771-788
• /
• 2019

Deep learning techniques such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) show superior performance in text classification than traditional approaches such as Support Vector Machines (SVMs) and Naïve Bayesian approaches. When using CNNs for text classification tasks, word embedding or character embedding is a step to transform words or characters to fixed size vectors before feeding them into convolutional layers. In this paper, we propose a parallel word-level and character-level embedding approach in CNNs for text classification. The proposed approach can capture word-level and character-level patterns concurrently in CNNs. To show the usefulness of proposed approach, we perform experiments with two English and three Korean text datasets. The experimental results show that character-level embedding works better in Korean and word-level embedding performs well in English. Also the experimental results reveal that the proposed approach provides better performance than traditional CNNs with word-level embedding or character-level embedding in both Korean and English documents. From more detail investigation, we find that the proposed approach tends to perform better when there is relatively small amount of data comparing to the traditional embedding approaches.

• Smart Media Journal
• /
• v.11 no.3
• /
• pp.26-30
• /
• 2022

Spiking neural network is a neural network that applies the working principle of real brain neurons. Due to the biological mechanism of neurons, it consumes less power for training and reasoning than conventional neural networks. Recently, as deep learning models become huge and operating costs increase exponentially, the spiking neural network is attracting attention as a third-generation neural network that connects convolution neural networks and recurrent neural networks, and related research is being actively conducted. However, in order to apply the spiking neural network model to the industry, a lot of research still needs to be done, and the problem of model retraining to apply a new model must also be solved. In this paper, we propose a method to minimize the cost of model retraining by extracting the weights of the existing trained deep learning model and converting them into the weights of the spiking neural network model. In addition, it was found that weight conversion worked correctly by comparing the results of inference using the converted weights with the results of the existing model.

• Journal of IKEEE
• /
• v.24 no.2
• /
• pp.523-528
• /
• 2020

With a widespread of sensor-rich mobile devices, the analysis of human activities becomes more general and simpler than ever before. In this paper, we propose two deep neural networks that efficiently and accurately perform human activity recognition (HAR) using tri-axial accelerometers. In combination with powerful modern deep learning techniques like batch normalization and LSTM networks, our model outperforms baseline approaches and establishes state-of-the-art results on WISDM dataset.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.11
• /
• pp.163-169
• /
• 1994

We propose a new learning method of recurrent neural networks as an effort to solve local minima problem. In this method the network with fixed connection weights is run for a given period time under given time-variant external inputs and initial conditions. The weights are changed in the direction that the total error is maximally decreased by using the steepest gradient method. If the obtained error is not sufficiently small even after iterating this procedure, additional feedback connections are introduced. Then, the external input signal is redefined. And we execute experiments on the recognition of Korean spoken digits as an application of the proposed network.

• Journal of the Korea Convergence Society
• /
• v.9 no.3
• /
• pp.15-21
• /
• 2018

The ocean drifter is a device for observing the ocean weather by floating off the sea surface. The data observed through the drifter is utilized in the ocean weather prediction and oil spill. Observed data may contain incorrect or missing data at the time of observation, and accuracy may be lowered when we use the data. In this paper, we propose a data correction model using recurrent neural networks. We corrected data collected from 7 drifters in 2015 and 8 drifters in 2016, and conducted experiments of drifter moving prediction to reflect the correction results. Experimental results showed that observed data are corrected by 13.9% and improved the performance of the prediction model by 1.4%.