• 응용통계연구
• /
• 제19권2호
• /
• pp.305-317
• /
• 2006

시계열 자료에서의 특이치, 특히 이 가운데 가법적 특이치가 모형의 식별, 모수의 추정 및 예측과 관련된 분석 전과정을 왜곡하는 것은 잘 알려져 있다. 그러나 특이치가 다수 발생하는 경우, 특히 연속적으로 집단을 이루어 발생할 때 대부분 특이치 검출방법은 가면화효과와 수렁화효과때문에 이들을 정확히 판별하지 못한다. 본 논문에서는 p차 자기상관회귀모형에 대한 고붕괴점 회귀추정량을 이용한 양방향 로버스트 필터방법을 제안했다. 실제 사례와 모의실험을 통해 제안한 방법이 매우 정확하게 시계열 자료에 포함된 특이치들을 검출하고 있음을 확인할 수 있다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제19권12호
• /
• pp.4260-4281
• /
• 2025

In image-based anomaly detection scenarios, current approaches such as deep neural network-based methods (DSVDD), self-supervised learning-based methods (DASVDD), and complete learning-based methods (CDSVDD), are primarily driven by positive (normal) samples. This reliance poses a risk wherein abnormal sample features may fall within the feature space of normal samples, especially when there is significant overlap between the distributions of normal and abnormal samples. To address the aforementioned issues, by absorbing the essence of outlier exposure open-set recognition, we propose an innovative method for anomaly detection called DSVDD-EOE. Our approach aims to minimize the enclosed hypersphere containing the feature region of positive samples while controlling the optimized edge outlier exposure set feature outside of this hypersphere. Unlike existing methods, the proposed method considers the center of the hypersphere as a learnable parameter that can be adjusted according to an evolved deep feature representation. In addition, we construct and optimize the outlier exposure set to participate in anomaly detection modeling, which significantly reduces the likelihood of mapping abnormal sample features into the domain of normal sample features. Experimental results demonstrate that the proposed method achieved an average area under the curve (AUC) value of 90.8% on the CIFAR-10 image benchmark dataset, which is 19.5% higher than that achieved by current state-of-the-art anomaly detection methods. On the FMNIST dataset, the proposed method achieved an impressive 95.8% average AUC value, and on the MNIST dataset, the proposed method achieved an average AUC value of 98.9%, both exceeding the performance of prior techniques. In addition, the proposed method demonstrated superior performance on the more demanding Tiny ImageNet dataset.

• Journal of applied mathematics & informatics
• /
• 제43권4호
• /
• pp.1141-1156
• /
• 2025

This research advances agricultural forecasting by integrating innovative statistical techniques, specifically Z-Score normalization and Winsorization, with Bayesian regression methodologies for paddy crop production prediction in Andhra Pradesh, India. By implementing Z-Score analysis to standardize data and Winsorization to mitigate extreme value impacts, the study develops a refined approach to posterior probability estimation using the inflection point methodology. The comprehensive methodological framework combines advanced preprocessing techniques with Bayesian Probabilistic Prediction Process, evaluating model performance through multiple statistical metrics including mean error, mean absolute error, and Theil's U statistic. The research demonstrates how sophisticated statistical preprocessing can enhance predictive accuracy, offering a nuanced approach to addressing data variability and outlier challenges in agricultural forecasting. By bridging advanced statistical methodologies with practical agricultural research, the study provides a robust framework for more precise and reliable crop output predictions, contributing to the evolving landscape of data-driven agricultural modeling.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제11권1호
• /
• pp.302-320
• /
• 2017

In this paper, an innovative robust feature detection and matching strategy for visual odometry based on stereo image sequence is proposed. First, a sparse multiscale 2D local invariant feature detection and description algorithm AKAZE is adopted to extract the interest points. A robust feature matching strategy is introduced to match AKAZE descriptors. In order to remove the outliers which are mismatched features or on dynamic objects, an improved random sample consensus outlier rejection scheme is presented. Thus the proposed method can be applied to dynamic environment. Then, geometric constraints are incorporated into the motion estimation without time-consuming 3-dimensional scene reconstruction. Last, an iterated sigma point Kalman Filter is adopted to refine the motion results. The presented ego-motion scheme is applied to benchmark datasets and compared with state-of-the-art approaches with data captured on campus in a considerably cluttered environment, where the superiorities are proved.

• 농업과학연구
• /
• 제42권1호
• /
• pp.53-61
• /
• 2015

Development of highly efficient and energy-saving tractors has been one of the issues in agricultural machinery. For design of such tractors, measurement and analysis of load on major power transmission parts of the tractors are the most important pre-requisite tasks. Objective of this study was to perform pre-processing procedures before effective analysis of load data of agricultural tractors (30, 75, and 82 kW) during major field operations such as plow tillage, rotary tillage, baling, bale wrapping, and to select the suitable pre-processing method for the analysis. A load measurement systems, equipped in the tractors, were consisted of strain-gauge, encoder, hydraulic pressure, and radar speed sensors to measure torque and rotational speed levels of transmission input shaft, PTO shaft, and driving axle shafts, pressure of the hydraulic inlet line, and travel speed, respectively. The entire sensor data were collected at a 200-Hz rate. Plow tillage, rotary tillage, baling, wrapping, and loader operations were selected as major field operations of agricultural tractors. Same or different farm works and driving levels were set differently for each of the load measuring experiment. Before load data analysis, pre-processing procedures such as outlier removal, low-pass filtering, and data division were performed. Data beyond the scope of the measuring range of the sensors and the operating range of the power transmission parts were removed. Considering engine and PTO rotational speeds, frequency components greater than 90, 60, and 60 Hz cut off frequencies were low-pass filtered for plow tillage, rotary tillage, and baler operations, respectively. Measured load data were divided into five parts: driving, working, implement up, implement down, and turning. Results of the study would provide useful information for load characteristics of tractors on major field operations.