• ETRI Journal
• /
• 제34권2호
• /
• pp.288-291
• /
• 2012

In this letter, a new online anomaly detection approach for software systems is proposed. The novelty of the proposed approach is to apply a new semantic kernel function for a support vector machine (SVM) classifier to detect fault-suspicious execution paths at runtime in a reasonable amount of time. The kernel uses a new sequence matching algorithm to measure similarities among program execution paths in a customized feature space whose dimensions represent the largest common subpaths among the execution paths. To increase the precision of the SVM classifier, each common subpath is given weights according to its ability to discern executions as correct or anomalous. Experiment results show that compared with the known kernels, the proposed SVM kernel will improve the time overhead of online anomaly detection by up to 170%, while improving the precision of anomaly alerts by up to 140%.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제15권3호
• /
• pp.145-152
• /
• 2015

We consider an online selective-sample learning problem for sequence classification, where the goal is to learn a predictive model using a stream of data samples whose class labels can be selectively queried by the algorithm. Given that there is a limit to the total number of queries permitted, the key issue is choosing the most informative and salient samples for their class labels to be queried. Recently, several aggressive selective-sample algorithms have been proposed under a linear model for static (non-sequential) binary classification. We extend the idea to hidden Markov models for multi-class sequence classification by introducing reasonable measures for the novelty and prediction confidence of the incoming sample with respect to the current model, on which the query decision is based. For several sequence classification datasets/tasks in online learning setups, we demonstrate the effectiveness of the proposed approach.

• International Journal of Computer Science & Network Security
• /
• 제21권6호
• /
• pp.77-88
• /
• 2021

Heart Failure represents a critical pathological case that is challenging to predict and discover at an early age, with a notable increase in morbidity and mortality. Machine Learning and Neural Network techniques play a crucial role in predicting heart attacks, diseases and more. These techniques give valuable perspectives for clinicians who may then adjust their diagnosis for each individual patient. This paper evaluated neural network models for heart attacks predictions. Several online learning methods were investigated to automatically and accurately predict heart attacks. The UCI dataset was used in this work to train and evaluate First Order and Second Order Online Learning methods; namely Backpropagation, Delta bar Delta, Levenberg Marquardt and QuickProp learning methods. An optimizer technique was also used to minimize the random noise in the database. A regularization concept was employed to further improve the generalization of the model. Results show that a three layers' NN model with a Backpropagation algorithm and Nadam optimizer achieved a promising accuracy for the heart attach prediction tasks.

• Journal of Information Technology Applications and Management
• /
• 제18권1호
• /
• pp.125-141
• /
• 2011

As the sales volume of online shops increases, the job burden in the back-offices of the online shops also increases. Order picking is the most labor-intensive operation among the jobs in a back-office and mid-size pure click online shops are experiencing the time delay and complexity in order picking nowadays while fulfilling their customers' orders. Those warehouses of the mid-size shops are based on manual systems, and as order pickings are repeated, the warehouses get a mess and lots of products in those warehouses are getting missing, which results in severe delay in order picking. To overcome this kind of problem in online clothing shops, we research a methodology to locate warehousing products. When products arrive at a warehouse, they are packed into a box and located on a rack in the warehouse. At this point, the operator should determine the box to be put in and the location on the rack for the box to be put on. This problem could be formulated as an Integer Programming model, but the branch-and bound algorithm to solve the IP model requires enormous computation, and sometimes it is even impossible to get a solution in a proper time. So, we relaxed the problem, developed a set of heuristics as a methodology to get a semi-optimum in an acceptable time, and proved by an experiment that the solutions by our methodology are satisfactory and acceptable by field managers.

• 한국시뮬레이션학회논문지
• /
• 제27권2호
• /
• pp.71-81
• /
• 2018

조선소의 지게차는 작업 특성상 무거운 물건을 상/하차하거나 이송하는 작업이 빈번하다. 작업은 동적이며 시간대별로 생성 비율이 다르다. 특히 오전과 오후 업무시간 직후에 작업 발생 비율이 높은 경향을 보인다. 이러한 상/하차 작업과 이송작업의 무게는 매번 다르며, 활용되는 지게차 역시 작업 가능한 허용무게의 제약이 있다. 본 연구에서는 지게차의 원활한 운영을 위해 최소 비용할당을 사용한 최근린 배차 규칙 알고리즘을 제안한다. 제시된 알고리즘은 다양한 종류의 지게차와 다수의 작업을 동시에 고려하여 배차를 결정하며, 지게차 종류에 따른 작업 불가능을 고려하기 위해 가상 지게차와 가상 작업을 생성하는 방법을 제안한다. 그리고 차량의 상태를 고려하여 체계적으로 지게차를 선택하는 방법도 함께 제시한다. 성능지표는 평균 공차이동거리와 평균 작업대기시간으로 한다. 성능비교를 위해 조선소의 지게차 운영방식을 모델링한 우선순위 규칙을 비교 대상으로 한다. 시뮬레이션을 통해 제시한 알고리즘의 우수성을 확인한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권2호
• /
• pp.710-726
• /
• 2018

This paper presents an extended particle filter to increase the accuracy and decrease the computation load of vehicle tracking. Particle filter has been the subject of extensive interest in video-based tracking which is capable of solving nonlinear and non-Gaussian problems. However, there still exist problems such as preventing unnecessary particle consumption, reducing the computational burden, and increasing the accuracy. We aim to increase the accuracy without an increase in computation load. In proposed method, we calculate the direction angle of the target vehicle. The angular difference between the direction of the target vehicle and each particle of the particle filter is observed. Particles are filtered and weighted, based on their angular difference. Particles with angular difference greater than a threshold is eliminated and the remaining are stored with greater weights in order to increase their probability for state estimation. Threshold value is very critical for performance. Thus, instead of having a constant threshold value, proposed algorithm updates it online. The first advantage of our algorithm is that it prevents the system from failures caused by insufficient amount of particles. Second advantage is to reduce the risk of using unnecessary number of particles in tracking which causes computation load. Proposed algorithm is compared against camshift, direction-based particle filter and condensation algorithms. Results show that the proposed algorithm outperforms the other methods in terms of accuracy, tracking duration and particle consumption.

• 한국통신학회논문지
• /
• 제30권12B
• /
• pp.811-817
• /
• 2005

유선 네트워크에 비해 상대적으로 제한된 대역폭을 가지는 셀룰러 네트워크의 특성으로 인해 효율적인 대역폭 관리에 대한 관심이 증가하고 있다. 또한, 최근 들어 수요가 증가하고 있는 다양한 멀티미디어 서비스들의 QoS를 보장하고, 사용자의 이동성으로 인해 발생할 수 있는 네트워크 과부하 현상을 해결하기 위한 실시간 대역폭 관리에 대한 중요성이 더욱 강조되고 있다 본 논문에서는 멀티미디어 셀룰러 네트워크상에서 대역폭의 이동을 통한 온라인 부하분산 알고리즘을 제안하였다 이 방법은 각 셀들간 트래픽 부하의 균형을 통해 지역적으로 발생하는 과부하 현상을 극복하고 높은 대역폭 효율성을 보장한다. 또한, 제안된 알고리즘은 현재 네트워크 상황에 대한 적응성과 유연성을 제공하는 온라인 기법을 기반으로 셀 단위로 수행되기 때문에 실제 네트워크 상황에 적용하기가 용이하다. 시뮬레이션을 통하여 제안된 방법이 기존의 타 기법들에 비해 네트워크의 다양한 트래픽 상황에서 우수한 성능을 가지는 것을 확인하였다.

• 한국의류학회지
• /
• 제45권5호
• /
• pp.892-906
• /
• 2021

This study explores the effects of providing information on the price fairness perception and intention of continuous use in an online fashion platform, given a price difference due to AI algorithm pricing. We investigated the moderating roles of price inequality (loss vs. gain) and technology insecurity. The experiments used four stimuli based on price inequality (loss vs. gain) and information provision (provided or not) on price inequality. We developed a mock website and offered a scenario on the product presentation based on an AI algorithm pricing. Participants in their 20s and 30s were randomly allocated to one of the stimuli. To test the hypotheses, a total of 257 responses were analyzed using Process Macro 3.4. According to the results, price fairness perception mediated between information provision and continuous use intention when consumers saw the price inequality as a gain. When the consumers perceived high technology insecurity, information provision affected the intention of continuous use mediated by price fairness perception.

• 자원환경지질
• /
• 제56권1호
• /
• pp.65-73
• /
• 2023

As one of the most vulnerable countries to floods, there should be an increased necessity for accurate and reliable flood forecasting in Indonesia. Therefore, a new prediction model using a machine learning algorithm is proposed to provide daily flood prediction in Indonesia. Data crawling was conducted to obtain daily rainfall, streamflow, land cover, and flood data from 2008 to 2021. The model was built using a Random Forest (RF) algorithm for classification to predict future floods by inputting three days of rainfall rate, forest ratio, and stream flow. The accuracy, specificity, precision, recall, and F1-score on the test dataset using the RF algorithm are approximately 94.93%, 68.24%, 94.34%, 99.97%, and 97.08%, respectively. Moreover, the AUC (Area Under the Curve) of the ROC (Receiver Operating Characteristics) curve results in 71%. The objective of this research is providing a model that predicts flood events accurately in Indonesian regions 3 months prior the day of flood. As a trial, we used the month of June 2022 and the model predicted the flood events accurately. The result of prediction is then published to the website as a warning system as a form of flood mitigation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권11호
• /
• pp.3050-3063
• /
• 2023

Recommendation systems provide personalized products or services to online users by mining their past preferences. Collaborative filtering is a popular recommendation technique because it is easy to implement. However, with the rapid growth of the number of users in recommendation systems, collaborative filtering suffers from serious scalability and sparsity problems. To address these problems, a novel collaborative filtering recommendation algorithm is proposed. The proposed algorithm partitions the users using affinity propagation clustering, and searches for k nearest neighbors in the partition where active user belongs, which can reduce the range of searching and improve real-time performance. When predicting the ratings of active user's unrated items, mean deviation method is used to impute values for neighbors' missing ratings, thus the sparsity can be decreased and the recommendation quality can be ensured. Experiments based on two different datasets show that the proposed algorithm is excellent both in terms of real-time performance and recommendation quality.