• Journal of the Korean Institute of Intelligent Systems
• /
• v.8 no.3
• /
• pp.9-15
• /
• 1998

This paper presents an adaptive learning control method for unknown nonlinear systems by combining neuro control and iterative learning control techniques. In the present control system, an iterative learning controller (ILC) is used for a process of short term memory involved in a temporary adaptive and learning manipulation and a short term storage of a specific temporary action. The learning gain of the iterative learning law is estimated by using a neural network for an unknown system except relative degrees. The control informations obtained by ILC are transferred to a long term memory-based feedforward neuro controller (FNC) and accumulated in it in addition to the previously stored infonnations. This scheme is applied to a two link robot manipulator through simulations.

• Proceedings of the Korea Contents Association Conference
• /
• 2017.05a
• /
• pp.421-422
• /
• 2017

콘텐츠 큐레이션 서비스를 위해서 대용량 데이터를 학습하는 과정에서 발생하는 메모리부족 문제, 학습소요시간 문제 등을 해결하기 위한 "대용량 문서학습을 위한 동적학습 파이프라인 생성기술 중 빅데이터 마이닝을 위한 점진적 학습 모델" 기술이 필요하며, 본 논문에서 제안한 콘텐츠 큐레이션 서비스는 온라인상의 수많은 콘텐츠들 중 개인의 주관이나 관점에 따라 관련 콘텐츠들을 수집, 정리하고 편집하여 이용자와 관련이 있거나 좋아할 만한 콘텐츠를 제공하는 서비스이다. 큐레이션 서비스에서는 개인비서, 금융 분야의 투자, 자율주행, 저널리즘, 효율적인 업무 지시/감독, 제조업의 자동화 공정, 교육, 콘텐츠 유통, 학술정보 등에서 컴퓨터가 방대한 양의 데이터로 부터 학습하여 사람의 일을 대신 처리하거나 의사결정에 도움을 줌으로써 업무의 효율을 높여주는 서비스 산업에 활용이 가능하다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2019.05a
• /
• pp.340-340
• /
• 2019

지구온난화로 유발되는 기후변동성이 증가함에 따라서 정확한 수문인자의 예측은 전 세계적으로 주요 관심사항이 되고 있다. 최근에는 고성능 컴퓨터 자원의 증가로 수문기상학 연구에서 동일한 학습량에 비하여 정확도의 향상이 뚜렷한 기계학습 구조를 활용하여 위성영상 기반의 대기예측, 태풍위치 추적 및 강수량 예측 등의 연구가 활발하게 진행되고 있다. 본 연구에는 기계학습 중 시계열 분석에 널리 활용되고 있는 순환신경망(Recurrent Neural Network, RNN) 기법의 대표적인 LSTM(Long Short-Term Memory) 네트워크를 이용하여 수문인자를 예측하였다. LSTM 네트워크는 가중치 및 메모리 요소에 대한 추가정보를 셀 상태에 저장하고 시계열의 길이 조정하여 모형의 탄력적 활용이 가능하다. LSTM 네트워크를 이용한 다양한 수문인자 예측결과 RMSE의 개선을 확인하였다. 따라서 본 연구를 통하여 개발된 기계학습을 통한 수문인자 예측기술은 권역별 수계별 홍수 및 가뭄대응 계획을 능동적으로 수립하는데 활용될 것으로 판단된다. 향후 연구에서는 LSTM의 입력영역을 Bayesian 추론기법을 활용하여 구성함으로 학습과정의 불확실성을 정량적으로 제어하고자 한다.

• Journal of Korea Foundry Society
• /
• v.41 no.6
• /
• pp.521-527
• /
• 2021

This study presents a process management method for the detection of casting defects during in high-pressure die casting based on machine learning. The model predicts the defects of the next cycle by extracting the features appearing over the previous cycles. For design of the gearbox, the proposed model detects shrinkage defects with data from three cycles in advance with 98.9% accuracy and 96.8% recall rates.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.19 no.11
• /
• pp.1254-1264
• /
• 2008

This paper suggests an extended memory polynomial model that improves accuracy in modeling memory effects of RF power amplifiers(PAs), and verifies effectiveness of the suggested method. The extended memory polynomial model includes cross-terms that are products of input terms that have different delay values to improve the limited accuracy of basic memory polynomial model that includes the diagonal terms of Volterra kernels. The complexity of the memoryless model, memory polynomial model, and the suggested model are compared. The extended memory polynomial model is represented with a matrix equation, and the Volterra kernels are extracted using least square method. In addition, the structure of digital predistorter and digital signal processing(DSP) algorithm based on the suggested model and indirect learning method are proposed to implement a digital predistortion linearization. To verify the suggested model, the predicted output of the model is compared with the measured output for a 10W GaN HEMT RF PA and 30 W LDMOS RF PA using 2.3 GHz WiBro input signal, and adjacent-channel power ratio(ACPR) performance with the proposed digital predistortion is measured. The proposed model increases model accuracy for the PAs, and improves the linearization performance by reducing ACPR.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.06a
• /
• pp.217-220
• /
• 2018

본 논문에서는 딥러닝을 활용한 비디오 보간법(video interpolation)에 대한 최근 모델들을 HD 급 비디오로 학습시키는 방법과 평가 성능을 비교 분석하는 것을 목표로 한다. 기존의 딥러닝을 활용한 비디오 보간법에 대해 제안된 모델들은 낮은 해상도의 비디오로 실험을 진행하였다. 반면 본 연구에서는 한정된 메모리를 가지고도 높은 해상도의 비디오를 학습시키기 위해서 패치 단위 데이터 셋을 구성하여 학습을 진행하였다. 평가 성능을 보이기 위해서 학습 데이터와 마찬가지로 패치 단위 평가와 전체 프레임 단위 평가 성능의 결과를 비교한다.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1055-1065
• /
• 2023

Recently, various studies using deep reinforcement learning (deep RL) technology have been conducted to solve complex problems using big data collected at industrial internet of things. Deep RL uses reinforcement learning"s trial-and-error algorithms and cumulative compensation functions to generate and learn its own data and quickly explore neural network structures and parameter decisions. However, studies so far have shown that the larger the size of the learning data is, the higher are the memory usage and search time, and the lower is the accuracy. In this study, model-agnostic learning for efficient federated deep RL was utilized to solve privacy invasion by increasing robustness as 55.9% and achieve 97.8% accuracy, an improvement of 5.5% compared with the comparative optimization-based meta learning models, and to reduce the delay time by 28.9% on average.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.05a
• /
• pp.249-251
• /
• 2021

In this paper, we introduce the deep-learning system using Tensorflow for recognizing situations that can occur fall situations when the elderly are moving or standing. Fall detection uses the LSTM (long short-term memory) learned using Tensorflow to determine whether it is a fall or not by data measured from wearable accelerator sensor. Learning is carried out for each of the 7 behavioral patterns consisting of 4 types of activity of daily living (ADL) and 3 types of fall. The learning was conducted using the 3-axis acceleration sensor data. As a result of the test, it was found to be compliant except for the GDSVM(Gravity Differential SVM), and it is expected that better results can be expected if the data is mixed and learned.

• Journal of Broadcast Engineering
• /
• v.27 no.6
• /
• pp.897-905
• /
• 2022

Research for video key frame detection has been actively conducted in the fields of computer vision. Recently with the advances on deep learning techniques, performance of key frame detection has been improved, but the various type of video content and complicated background are still a problem for efficient learning. In this paper, we propose a novel method for key frame detection, witch utilizes contrastive learning and memory bank module. The proposed method trains the feature extracting network based on the difference between neighboring frames and frames from separate videos. Founded on the contrastive learning, the method saves and updates key frames in the memory bank, witch efficiently reduce redundancy from the video. Experimental results on video dataset show the effectiveness of the proposed method for key frame detection.

• Journal of KIISE
• /
• v.44 no.1
• /
• pp.51-56
• /
• 2017

The research area for solving question answering (QA) problems using artificial intelligence models is in a methodological transition period, and one such architecture, the dynamic memory network (DMN), is drawing attention for two key attributes: its attention mechanism defined by neural network operations and its modular architecture imitating cognition processes during QA of human. In this paper, we increased accuracy of the inferred answers, by adapting an automatic data augmentation method for lacking amount of training data, and by improving the ability of time perception. The experimental results showed that in the 1K-bAbI tasks, the modified DMN achieves 89.21% accuracy and passes twelve tasks which is 13.58% higher with passing four more tasks, as compared with one implementation of DMN. Additionally, DMN's word embedding vectors form strong clusters after training. Moreover, the number of episodic passes and that of supporting facts shows direct correlation, which affects the performance significantly.