• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.1
• /
• pp.123-130
• /
• 2021

With the recent developments in computer technology, there has been an increasing interest in the field of machine learning. This also has led to a significant increase in real business cases of machine learning theory in various sectors. In finance, it has been a major challenge to predict the future value of financial products. Since the 1980s, the finance industry has relied on technical and fundamental analysis for this prediction. For future value prediction models using machine learning, model design is of paramount importance to respond to market variables. Therefore, this paper quantitatively predicts the stock price movements of individual stocks listed on the KOSPI market using machine learning techniques; specifically, the reinforcement learning model. The DQN and A2C algorithms proposed by Google Deep Mind in 2013 are used for the reinforcement learning and they are applied to the stock trading strategies. In addition, through experiments, an input value to increase the cumulative profit is selected and its superiority is verified by comparison with comparative algorithms.

• Journal of The Korean Association of Information Education
• /
• v.25 no.3
• /
• pp.571-578
• /
• 2021

This study compares Online SW-based maker education in terms of self-efficacy, learning motivation, and interest after applying differently according to blended learning strategies. First, a SW maker program for blended learning was developed and applied as a live seminar-type class including real-time interactive and a support-providing class consisting of online content and Q&A. As a result of comparing the differences between students according to the two strategies divided into pre- and post- survey, in the self-efficacy part, there was a significant difference in the positive efficacy and the overall part, and in the learning motivation part, the live seminar form was significantly higher in the confidence part. In the interest part, the support-providing form showed a significantly higher average in the instrumental interest and nervous part. In order to maintain the effect of maker activities like existing face-to-face situations in Online learning, it is necessary to increase sharing time between students, an integrated learning environment, and sufficient provision of exploration time and learning materials.

• Journal of Digital Convergence
• /
• v.19 no.12
• /
• pp.347-352
• /
• 2021

In this paper, we propose a method for finding feature subsets that are effective for classification in an input dataset by using a multi-agent reinforcement learning method. In the field of machine learning, it is crucial to find features suitable for classification. A dataset may have numerous features; while some features may be effective for classification or prediction, others may have little or rather negative effects on results. In machine learning problems, feature selection for increasing classification or prediction accuracy is a critical problem. To solve this problem, we proposed a feature selection method based on reinforced learning. Each feature has one agent, which determines whether the feature is selected. After obtaining corresponding rewards for each feature that is selected, but not by the agents, the Q-value of each agent is updated by comparing the rewards. The reward comparison of the two subsets helps agents determine whether their actions were right. These processes are performed as many times as the number of episodes, and finally, features are selected. As a result of applying this method to the Wisconsin Breast Cancer, Spambase, Musk, and Colon Cancer datasets, accuracy improvements of 0.0385, 0.0904, 0.1252 and 0.2055 were shown, respectively, and finally, classification accuracies of 0.9789, 0.9311, 0.9691 and 0.9474 were achieved, respectively. It was proved that our proposed method could properly select features that were effective for classification and increase classification accuracy.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.723-724
• /
• 2022

지난 코로나 상황 동안 비대면 수업을 진행했고, 학생들은 빠르게 적응했다. 온라인 수업은 학습자가 이해될 때까지 반복 학습이 가능하고, 시간과 공간의 제약 없이 자기 주도적으로 학습할 수 있다는 장점이 있지만, 온라인상이라는 특징 때문에 교수자와 학습자 간 상호작용이 부족하다는 한계점이 존재한다. 하지만 이점은 차후에 비대면 수업의 지속적인 활용 및 확대를 제한하는 요인이 될 수 있다. 본 논문에서는 상호작용을 높일 수 있는 웹 기반 에듀테크 시스템을 제안한다. 온라인 수업의 강의 영상을 세부적인 내용을 나누는 Section을 통해 다른 학생들이 질문했던 Q&A 데이터를 모아서 생성된 Section-FAQ를 열람할 수 있고, 그 Q&A에 반응해서 상호작용이 가능하다. 또한 교수자에게 Q&A를 보낼 때 영상의 Section 정보와 강의시간 정보를 같이 전송하여 강의 영상을 확인하지 않고, 빠른 답변이 가능하도록 설계했다. 본 논문에서 제안하는 온라인 수업의 상호작용 향상을 위한 웹 기반 에듀테크 시스템을 통해 온라인상에서 교수자의 역할을 대신해주고 비대면 수업의 단점을 해소해주면서, 교수자과 학습자 간의 상호작용을 높여 수업의 이해도를 높이고 학습자들의 학업성취를 높일 수 있을 것이다.

• Annual Conference of KIPS
• /
• 2002.04a
• /
• pp.565-568
• /
• 2002

본 논문에서는 에이전트간의 통신이 불가능한 다중 에이전트 환경에서 각 에이전트들이 독립적이면서 대표적인 강화학습법인 Q-학습을 전개함으로써 서로 효과적으로 협조할 수 있는 행동전략을 학습하려고 한다. 하지만 단일 에이전트 경우에 비해 보다 큰 상태-행동공간을 갖는 다중 에이전트환경에서는 강화학습을 통해 효과적으로 최적의 행동 전략에 도달하기 어렵다는 문제점이 있다. 이 문제에 대한 기존의 접근방법은 크게 모듈화 방법과 일반화 방법이 제안되었으나 모두 나름의 제한을 가지고 있다. 본 논문에서는 대표적인 다중 에이전트 학습 문제의 예로서 the Prey and Hunters Problem를 소개하고 이 문제영역을 통해 이와 같은 강화학습의 문제점을 살펴보고, 해결책으로 신경망 SOM 을 이용한 일반화 방법을 제안한다. 이 방법은 다층 퍼셉트론 신경망과 역전파 알고리즘을 이용한 기존의 일반화 방법과는 달리 군집화 기능을 제공하는 신경망 SOM 을 이용함으로써 명확한 다수의 훈련 예가 없어도 효과적으로 채 경험하지 못한 상태-행동들에 대한 Q 값을 예측하고 이용할 수 있다는 장점이 있다.

• Korean Linguistics
• /
• v.80
• /
• pp.151-175
• /
• 2018

This paper proposed a method of semantic recognition to improve the extraction of correct answers of the Q&A system through machine learning. For this purpose, the semantic recognition method is described based on the distribution of predicative nouns. Predicative noun vocabularies and sentences were collected from Wikipedia documents. The predicative nouns are typed by analyzing the environment in which the predicative nouns appear in sentences. This paper proposes a semantic recognition method of predicative nouns to which rules can be applied. In Chapter 2, previous studies on predicative nouns were reviewed. Chapter 3 explains how predicative nouns are distributed. In this paper, every predicative nouns that can not be processed by rules are excluded, therefore, the predicative nouns noun forms combined with the case marker '의' were excluded. In Chapter 4, we extracted 728 sentences composed of 10,575 words from Wikipedia. A semantic analysis engine tool of ETRI was used and presented a predicative nouns noun that can be handled semantic recognition language.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.375-377
• /
• 2024

본 논문에서는 OpenAI Gym 환경에서 프로그램으로 간단한 제어가 가능한 Mountain-Car-v0 게임에 대해 DQN(Deep Q-Networks) 강화학습을 진행하였다. 본 논문에서 적용한 DQN 네트워크는 입력층 1개, 은닉층 3개, 출력층 1개로 구성하였고, 입력층과 은닉층에서의 활성화함수는 ReLU를, 출력층에서는 Linear함수를 활성화함수로 적용하였다. 실험은 Mountain-Car-v0에 대해 DQN 강화학습을 진행했을 때 각 에피소드별로 획득한 보상 결과를 살펴보고, 보상구간에 포함된 횟수를 분석하였다. 실험결과 전체 100회의 에피소드 중 보상을 50 이상 획득한 에피소드가 85개로 나타났다.

• The Journal of Korea Robotics Society
• /
• v.12 no.3
• /
• pp.297-305
• /
• 2017

As the development of autonomous vehicles becomes realistic, many automobile manufacturers and components producers aim to develop 'completely autonomous driving'. ADAS (Advanced Driver Assistance Systems) which has been applied in automobile recently, supports the driver in controlling lane maintenance, speed and direction in a single lane based on limited road environment. Although technologies of obstacles avoidance on the obstacle environment have been developed, they concentrates on simple obstacle avoidances, not considering the control of the actual vehicle in the real situation which makes drivers feel unsafe from the sudden change of the wheel and the speed of the vehicle. In order to develop the 'completely autonomous driving' automobile which perceives the surrounding environment by itself and operates, ability of the vehicle should be enhanced in a way human driver does. In this sense, this paper intends to establish a strategy with which autonomous vehicles behave human-friendly based on vehicle dynamics through the reinforcement learning that is based on Q-learning, a type of machine learning. The obstacle avoidance reinforcement learning proceeded in 5 simulations. The reward rule has been set in the experiment so that the car can learn by itself with recurring events, allowing the experiment to have the similar environment to the one when humans drive. Driving Simulator has been used to verify results of the reinforcement learning. The ultimate goal of this study is to enable autonomous vehicles avoid obstacles in a human-friendly way when obstacles appear in their sight, using controlling methods that have previously been learned in various conditions through the reinforcement learning.

• Journal of Korean Academy of Nursing
• /
• v.34 no.6
• /
• pp.994-1003
• /
• 2004

Purpose: This study was to develop Web-based multimedia content for Physical Examination and Health Assesment. Method: The multimedia content was developed based on Jung's teaching and learning structure plan model, using the following 5 processes: 1) Analysis Stage, 2) Planning Stage, 3) Storyboard Framing and Production Stage, 4) Program Operation Stage, and 5) Final Evaluation Stage. Results: The web based multimedia content consisted of an intro movie, main page and sub pages. On the main page, there were 6 menu bars that consisted of Announcement center, Information of professors, Lecture guide, Cyber lecture, Q&A, and Data centers, and a site map which introduced 15 week lectures. In the operation of web based multimedia content, HTML, JavaScript, Flash, and multimedia technology(Audio and Video) were utilized and the content consisted of text content, interactive content, animation, and audio & video. Consultation with the experts in context, computer engineering, and educational technology was utilized in the development of these processes. Conclusions: Web-based multimedia content is expected to offer individualized and tailored learning opportunities to maximize and facilitate the effectiveness of the teaching and learning process. Therefore, multimedia content should be utilized concurrently with the lecture in the Physical Examination and Health Assesment classes as a vital teaching aid to make up for the weakness of the face-to- face teaching-learning method.

• Proceedings of the KIEE Conference
• /
• 2009.07a
• /
• pp.778_779
• /
• 2009

The maximum output torque developed by the machine is dependent on the allowable current rating and maximum voltage that the inverter can supply to the machine. Therefore, to use the inverter capacity fully, it is desirable to use the control scheme considering the voltage and current limit condition, which can yield the maximum torque per ampere over the entire speed range. The paper is proposed maximum torque control of induction motor drive using adaptive learning neuro fuzzy controller and artificial neural network(ANN). The control method is applicable over the entire speed range and considered the limits of the inverter's current and voltage rated value. For each control mode, a condition that determines the optimal d, q axis current $_i_{ds}$, $i_{qs}$ for maximum torque operation is derived. The proposed control algorithm is applied to induction motor drive system controlled adaptive learning neuro fuzzy controller and ANN controller, the operating characteristics controlled by maximum torque control are examined in detail. Also, this paper is proposed the analysis results to verify the effectiveness of the adaptive learning neuro fuzzy controller and ANN controller.