• Journal of Information Processing Systems
• /
• v.14 no.1
• /
• pp.150-161
• /
• 2018

We propose a novel deep neural network model for detecting human activities in untrimmed videos. The process of human activity detection in a video involves two steps: a step to extract features that are effective in recognizing human activities in a long untrimmed video, followed by a step to detect human activities from those extracted features. To extract the rich features from video segments that could express unique patterns for each activity, we employ two different convolutional neural network models, C3D and I-ResNet. For detecting human activities from the sequence of extracted feature vectors, we use BLSTM, a bi-directional recurrent neural network model. By conducting experiments with ActivityNet 200, a large-scale benchmark dataset, we show the high performance of the proposed DeepAct model.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.4
• /
• pp.109-118
• /
• 2022

The data of the missing section among the vertex surface sea temperature observation data was imputed using the Bidirectional Recurrent Neural Network(BiRNN). Among artificial intelligence techniques, Recurrent Neural Networks (RNNs), which are commonly used for time series data, only estimate in the direction of time flow or in the reverse direction to the missing estimation position, so the estimation performance is poor in the long-term missing section. On the other hand, in this study, estimation performance can be improved even for long-term missing data by estimating in both directions before and after the missing section. Also, by using all available data around the observation point (sea surface temperature, temperature, wind field, atmospheric pressure, humidity), the imputation performance was further improved by estimating the imputation data from these correlations together. For performance verification, a statistical model, Multivariate Imputation by Chained Equations (MICE), a machine learning-based Random Forest model, and an RNN model using Long Short-Term Memory (LSTM) were compared. For imputation of long-term missing for 7 days, the average accuracy of the BiRNN/statistical models is 70.8%/61.2%, respectively, and the average error is 0.28 degrees/0.44 degrees, respectively, so the BiRNN model performs better than other models. By applying a temporal decay factor representing the missing pattern, it is judged that the BiRNN technique has better imputation performance than the existing method as the missing section becomes longer.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.12 no.2
• /
• pp.121-130
• /
• 2012

This paper proposes an intuitive real-time robot control system using human body movement. Recently, it has been developed that motion generation for humanoid robots with reflecting human body movement, which is measured by a motion capture. However, in the existing studies about robot control system by human body movement, the detailed structure information of a robot, for example, degrees of freedom, the range of motion and forms, must be examined in order to calculate inverse kinematics. In this study, we have proposed Associative Motion Generation as humanoid robot motion generation method which does not need the detailed structure information. The associative motion generation system is composed of two neural networks: nonlinear principal component analysis and Jordan recurrent neural network, and the associative motion is generated with the following three steps. First, the system learns the correspondence relationship between an indication and a motion using training data. Second, associative values are extracted for associating a new motion from an unfamiliar indication using nonlinear principal component analysis. Last, the robot generates a new motion through calculation by Jordan recurrent neural network using the associative values. In this paper, we propose a real-time humanoid robot control system based on Associative Motion Generation, that enables user to control motion intuitively by human body movement. Through the task processing and subjective evaluation experiments, we confirmed the effective usability and affective evaluations of the proposed system.

• Journal of Power Electronics
• /
• v.11 no.5
• /
• pp.719-725
• /
• 2011

This paper describes a novel Direct Torque Control (DTC) method for adjustable speed Doubly-Fed Induction Machine (DFIM) drives which is supplied by a two-level Space Vector Modulation (SVM) voltage source inverter (DTC-SVM) in the rotor circuit. The inverter reference voltage vector is obtained by using input-output feedback linearization control and a DFIM model in the stator a-b axes reference frame with stator currents and rotor fluxes as state variables. Moreover, to make this nonlinear controller stable and robust to most varying electrical parameter uncertainties, a two layer recurrent Artificial Neural Network (ANN) is used to estimate a certain function which shows the machine lumped uncertainty. The overall system stability is proved by the Lyapunov theorem. It is shown that the torque and flux tracking errors as well as the updated weights of the ANN are uniformly ultimately bounded. Finally, effectiveness of the proposed control approach is shown by computer simulation results.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.31 no.3
• /
• pp.401-410
• /
• 2021

Recently, as cyber attacks targeting industrial control systems increase, various studies are being conducted on the detection of abnormalities in industrial processes. Considering that the industrial process is deterministic and regular, It is appropriate to determine abnormality by comparing the predicted value of the detection model from which normal data is trained and the actual value. In this paper, HAI Datasets 20.07 and 21.03 are used. In addition, an ensemble model is created by combining models that have applied different time steps to Gated Recurrent Units. Then, the detection performance of the single model and the ensemble recurrent neural networks model were compared through various performance evaluation analysis, and It was confirmed that the proposed model is more suitable for abnormal detection in industrial control systems.

• Journal of Institute of Control, Robotics and Systems
• /
• v.6 no.6
• /
• pp.441-446
• /
• 2000

This paper presents an approach for the prediction of optimal heating load using a diagonal recurrent neural networks(DRNN) and data base system of outdoor temperature. In the DRNN, a dynamic backpropagation(DBP) with delta-bar-delta teaming method is used to train an optimal heating load identifier. And the data base system is utilized for outdoor temperature prediction. Compared to other kinds of methods, the proposed method gives better prediction performance of heating load. Also a hardware for the controller is developed using a microprocessor. The experimental results show that prediction enhancement for heating load can be achieved with the proposed method regardless of the its inherent nonlinearity and large time constant.

• Journal of Broadcast Engineering
• /
• v.24 no.3
• /
• pp.495-505
• /
• 2019

Typical algorithms of deep learning include CNN(Convolutional Neural Networks), which are mainly used for image recognition, and RNN(Recurrent Neural Networks), which are used mainly for speech recognition and natural language processing. Among them, CNN is able to learn from filters that generate feature maps with algorithms that automatically learn features from data, making it mainstream with excellent performance in image recognition. Since then, various algorithms such as R-CNN and others have appeared in object detection to improve performance of CNN, and algorithms such as YOLO(You Only Look Once) and SSD(Single Shot Multi-box Detector) have been proposed recently. However, since these deep learning-based detection algorithms determine the success of the detection in the still images, stable object tracking and detection in the video requires separate tracking capabilities. Therefore, this paper proposes a method of combining Kalman filters into deep learning-based detection networks for improved object tracking and detection performance in the video. The detection network used YOLO v2, which is capable of real-time processing, and the proposed method resulted in 7.7% IoU performance improvement over the existing YOLO v2 network and 20 fps processing speed in FHD images.

• Journal of Institute of Control, Robotics and Systems
• /
• v.14 no.3
• /
• pp.254-262
• /
• 2008

In this paper, we propose an adaptive neural dynamic surface control (DSC) approach with $H_{\infty}$ tracking performance for full dynamics of nonlinear flight systems. It is assumed that the model uncertainties such as structured and unstrutured uncertainties, and external disturbances influence the nonlinear aircraft model. In our control system, self recurrent wavelet neural networks (SRWNNs) are used to compensate the model uncertainties of nonlinear flight systems, and an adaptive DSC technique is extended for the disturbance attenuation of nonlinear flight systems. All weights of SRWNNs are trained on-line by the smooth projection algorithm. From Lyapunov stability theorem, it is shown that $H_{\infty}$ performance nom external disturbances can be obtained. Finally, we present the simulation results for a nonlinear six-degree-of-freedom F-16 aircraft model to confirm the effectiveness of the proposed control system.

• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1290-1295
• /
• 2017

Document classification, a task of classifying the category of each document based on text, is one of the fundamental areas for natural language processing. Document classification may be used in various fields such as topic classification and sentiment classification. Neural network models for document classification can be divided into two categories: word-level models and character-level models that treat words and characters as basic units respectively. In this study, we propose a neural network model that combines character-level and word-level models to improve performance of document classification. The proposed model extracts the feature vector of each word by combining information obtained from a word embedding matrix and information encoded by a character-level neural network. Based on feature vectors of words, the model classifies documents with a hierarchical structure wherein recurrent neural networks with attention mechanisms are used for both the word and the sentence levels. Experiments on real life datasets demonstrate effectiveness of our proposed model.

• Journal of Korea Multimedia Society
• /
• v.25 no.11
• /
• pp.1643-1652
• /
• 2022

Most of the existing machine reading research has used Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) algorithms as networks. Among them, RNN was slow in training, and Question Answering Network (QANet) was announced to improve training speed. QANet is a model composed of CNN and self-attention. CNN extracts semantic and syntactic information well from the local corpus, but there is a limit to extracting the corresponding information from the global corpus. Graph Convolutional Networks (GCN) extracts semantic and syntactic information relatively well from the global corpus. In this paper, to take advantage of this strength of GCN, we propose I-QANet, which changed the CNN of QANet to GCN. The proposed model performed 1.2 times faster than the baseline in the Stanford Question Answering Dataset (SQuAD) dataset and showed 0.2% higher performance in Exact Match (EM) and 0.7% higher in F1. Furthermore, in the Korean Question Answering Dataset (KorQuAD) dataset consisting only of Korean, the learning time was 1.1 times faster than the baseline, and the EM and F1 performance were also 0.9% and 0.7% higher, respectively.