• Electronics and Telecommunications Trends
• /
• v.33 no.4
• /
• pp.43-53
• /
• 2018

As face recognition (FR) has been well studied over the past decades, FR technology has been applied to many real-world applications such as surveillance and biometric systems. However, in the real-world scenarios, FR performances have been known to be significantly degraded owing to variations in face images, such as the pose, illumination, and low-resolution. Recently, visual intelligence technology has been rapidly growing owing to advances in deep learning, which has also improved the FR performance. Furthermore, the FR performance based on deep learning has been reported to surpass the performance level of human perception. In this article, we discuss deep-learning based high-performance FR technologies in terms of representative deep-learning based FR architectures and recent FR algorithms robust to face image variations (i.e., pose-robust FR, illumination-robust FR, and video FR). In addition, we investigate big face image datasets widely adopted for performance evaluations of the most recent deep-learning based FR algorithms.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.489-489
• /
• 2000

Recently, Reinforcement Learning(RL) methods have been used far teaming problems in Partially Observable Markov Decision Process(POMDP) environments. Conventional RL-methods, however, have limited applicability to POMDP To overcome the partial observability, several algorithms were proposed [5], [7]. The aim of this paper is to extend our previous algorithm for POMDP, called Labeling Q-learning(LQ-learning), which reinforces incomplete information of perception with labeling. Namely, in the LQ-learning, the agent percepts the current states by pair of observation and its label, and the agent can distinguish states, which look as same, more exactly. Labeling is carried out by a hash-like function, which we call Labeling Function(LF). Numerous labeling functions can be considered, but in this paper, we will introduce several labeling functions based on only 2 or 3 immediate past sequential observations. We introduce the basic idea of LQ-learning briefly, apply it to maze problems, simple POMDP environments, and show its availability with empirical results, look better than conventional RL algorithms.

• Proceedings of the KSME Conference
• /
• 2001.06b
• /
• pp.724-729
• /
• 2001

This paper discusses the composition of the theory of reinforcement learning, which is applied in real-time learning, and evolutionary strategy, which proves its the superiority in the finding of the optimal solution at the off-line learning method. The individuals are reduced in order to learn the evolutionary strategy in real-time, and new method that guarantee the convergence of evolutionary mutations are proposed. It possible to control the control object varied as time changes. As the state value of the control object is generated, applied evolutionary strategy each sampling time because the learning process of an estimation, selection, mutation in real-time. These algorithms can be applied, the people who do not have knowledge about the technical tuning of dynamic systems could design the controller or problems in which the characteristics of the system dynamics are slightly varied as time changes. In the future, studies are needed on the proof of the theory through experiments and the characteristic considerations of the robustness against the outside disturbances.

• ETRI Journal
• /
• v.41 no.5
• /
• pp.560-573
• /
• 2019

Two supervised learning algorithms, a basic neural network and a long short-term memory recurrent neural network, are applied to traffic including DDoS attacks. The joint effects of preprocessing methods and hyperparameters for machine learning on performance are investigated. Values representing attack characteristics are extracted from datasets and preprocessed by two methods. Binary classification and two optimizers are used. Some hyperparameters are obtained exhaustively for fast and accurate detection, while others are fixed with constants to account for performance and data characteristics. An experiment is performed via TensorFlow on three traffic datasets. Three scenarios are considered to investigate the effects of learning former traffic on sequential traffic analysis and the effects of learning one dataset on application to another dataset, and determine whether the algorithms can be used for recent attack traffic. Experimental results show that the used preprocessing methods, neural network architectures and hyperparameters, and the optimizers are appropriate for DDoS attack detection. The obtained results provide a criterion for the detection accuracy of attacks.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.11
• /
• pp.2897-2907
• /
• 1998

The production-rule generation from training examples is a hard problem that has large space and many local optimal solutions. Many learning methods are proposed for production-rule generation and genetic algorithms is an alternative learning method. However, traditional genetic algorithms has been known to have an obstacle in converging at the global solution area and show poor efficiency of production-rules generated. In this paper, we propose a production-rule generating method which uses genetic algorithm learning. By analyzing optimal sub-solutions captured by genetic algorithm learning, our method takes advantage of its schema structure and thus generates relatively small rule set.

• Journal of Korean Institute of Industrial Engineers
• /
• v.39 no.1
• /
• pp.30-45
• /
• 2013

Predicting monthly electricity price has been a significant factor of decision-making for plant resource management, fuel purchase plan, plans to plant, operating plan budget, and so on. In this paper, we propose a sophisticated prediction model in terms of the technique of modeling and the variety of the collected variables. The proposed model hybridizes the semi-supervised learning and the artificial neural network algorithms. The former is the most recent and a spotlighted algorithm in data mining and machine learning fields, and the latter is known as one of the well-established algorithms in the fields. Diverse economic/financial indexes such as the crude oil prices, LNG prices, exchange rates, composite indexes of representative global stock markets, etc. are collected and used for the semi-supervised learning which predicts the up-down movement of the price. Whereas various climatic indexes such as temperature, rainfall, sunlight, air pressure, etc, are used for the artificial neural network which predicts the real-values of the price. The resulting values are hybridized in the proposed model. The excellency of the model was empirically verified with the monthly data of electricity price provided by the Korea Energy Economics Institute.

• Industrial Engineering and Management Systems
• /
• v.8 no.3
• /
• pp.171-180
• /
• 2009

Population learning algorithm (PLA) is a population-based method that was inspired by the similarities to the phenomenon of social education process in which a diminishing number of individuals enter an increasing number of learning stages. The study aims to develop a framework that repeatedly applying the PLA to solve the discrete resource constrained project scheduling problem with two objectives: minimizing project makespan and renewable resource availability, which are two most common concerns of management when a project is being executed. The PLA looping framework will provide a number of near Pareto optimal schedules for the management to make a choice. Different improvement schemes and learning procedures are applied at different stages of the process. The process gradually becomes more and more sophisticated and time consuming as there are less and less individuals to be taught. An experiment with ProGen generated instances was conducted, and the results demonstrated that the looping framework using PLA outperforms those using genetic local search, particle swarm optimization with local search, scatter search, as well as biased sampling multi-pass algorithm, in terms of several performance measures of proximity. However, the diversity using spread metric does not reveal any significant difference between these five looping algorithms.

• The Journal of Information Systems
• /
• v.29 no.3
• /
• pp.119-144
• /
• 2020

Purpose Real estate usually takes charge of the highest proportion of physical properties which individual, organizations, and government hold and instability of real estate market affects the economic condition seriously for each economic subject. Consequently, practices for predicting the real estate market have attention for various reasons, such as financial investment, administrative convenience, and wealth management. Additionally, development of machine learning algorithms and computing hardware enhances the expectation for more precise and useful prediction models in real estate market. Design/methodology/approach In response to the demand, this paper aims to provide a framework for forecasting the real estate market with machine learning algorithms. The framework consists of demonstrating the prediction efficiency of each machine learning algorithm, interpreting the interior feature effects of prediction model with a state-of-art algorithm, LIME(Local Interpretable Model-agnostic Explanation), and comparing the results in different cities. Findings This research could not only enhance the academic base for information system and real estate fields, but also resolve information asymmetry on real estate market among economic subjects. This research revealed that macroeconomic indicators, real estate-related indicators, and Google Trends search indexes can predict real-estate prices quite well.

• Journal of the Korea Safety Management & Science
• /
• v.23 no.4
• /
• pp.93-104
• /
• 2021

In this paper, we propose a method for diagnosing overload and working load of collaborative robots through performance analysis of machine learning algorithms. To this end, an experiment was conducted to perform pick & place operation while changing the payload weight of a cooperative robot with a payload capacity of 10 kg. In this experiment, motor torque, position, and speed data generated from the robot controller were collected, and as a result of t-test and f-test, different characteristics were found for each weight based on a payload of 10 kg. In addition, to predict overload and working load from the collected data, machine learning algorithms such as Neural Network, Decision Tree, Random Forest, and Gradient Boosting models were used for experiments. As a result of the experiment, the neural network with more than 99.6% of explanatory power showed the best performance in prediction and classification. The practical contribution of the proposed study is that it suggests a method to collect data required for analysis from the robot without attaching additional sensors to the collaborative robot and the usefulness of a machine learning algorithm for diagnosing robot overload and working load.

• Journal of Platform Technology
• /
• v.11 no.4
• /
• pp.80-88
• /
• 2023

To construct a learning model in the field of artificial intelligence, a dataset should be collected and be delivered to the central server where the learning model is constructed. Federated learning is a machine learning method building a global learning model without transmitting data located in a client side in a collaborative manner. It can be used to protect privacy, and after constructing a local trained model on individual clients, the parameters of the local model are aggregated centrally to update the global model. In this paper, we reuse the existing learning parameter to improve federated learning, describe incremental federated learning. For this work, we do experiments using the federated learning framework named Flower, and evaluate the experiment results with regard to elapsed time and precision when executing optimization algorithms.