• Title/Summary/Keyword: Feature selection

Search Result 1,076, Processing Time 0.028 seconds

Fault Classification for Rotating Machinery Using Support Vector Machines with Optimal Features Corresponding to Each Fault Type (결함유형별 최적 특징과 Support Vector Machine 을 이용한 회전기계 결함 분류)

  • Kim, Yang-Seok;Lee, Do-Hwan;Kim, Seong-Kook
    • Transactions of the Korean Society of Mechanical Engineers A
    • /
    • v.34 no.11
    • /
    • pp.1681-1689
    • /
    • 2010
  • Several studies on the use of Support Vector Machines (SVMs) for diagnosing rotating machinery have been successfully carried out, but the fault classification depends on the input features as well as a multi-classification scheme, binary optimizer, kernel function, and the parameter to be used in the kernel function. Most of the published papers on multiclass SVM applications report the use of the same features to classify the faults. In this study, simple statistical features are determined on the basis of time domain vibration signals for various fault conditions, and the optimal features for each fault condition are selected. Then, the optimal features are used in the SVM training and in the classification of each fault condition. Simulation results using experimental data show that the results of the proposed stepwise classification approach with a relatively short training time are comparable to those for a single multi-class SVM.

Runtime Prediction Based on Workload-Aware Clustering (병렬 프로그램 로그 군집화 기반 작업 실행 시간 예측모형 연구)

  • Kim, Eunhye;Park, Ju-Won
    • Journal of Korean Society of Industrial and Systems Engineering
    • /
    • v.38 no.3
    • /
    • pp.56-63
    • /
    • 2015
  • Several fields of science have demanded large-scale workflow support, which requires thousands of CPU cores or more. In order to support such large-scale scientific workflows, high capacity parallel systems such as supercomputers are widely used. In order to increase the utilization of these systems, most schedulers use backfilling policy: Small jobs are moved ahead to fill in holes in the schedule when large jobs do not delay. Since an estimate of the runtime is necessary for backfilling, most parallel systems use user's estimated runtime. However, it is found to be extremely inaccurate because users overestimate their jobs. Therefore, in this paper, we propose a novel system for the runtime prediction based on workload-aware clustering with the goal of improving prediction performance. The proposed method for runtime prediction of parallel applications consists of three main phases. First, a feature selection based on factor analysis is performed to identify important input features. Then, it performs a clustering analysis of history data based on self-organizing map which is followed by hierarchical clustering for finding the clustering boundaries from the weight vectors. Finally, prediction models are constructed using support vector regression with the clustered workload data. Multiple prediction models for each clustered data pattern can reduce the error rate compared with a single model for the whole data pattern. In the experiments, we use workload logs on parallel systems (i.e., iPSC, LANL-CM5, SDSC-Par95, SDSC-Par96, and CTC-SP2) to evaluate the effectiveness of our approach. Comparing with other techniques, experimental results show that the proposed method improves the accuracy up to 69.08%.

Offline Based Ransomware Detection and Analysis Method using Dynamic API Calls Flow Graph (다이나믹 API 호출 흐름 그래프를 이용한 오프라인 기반 랜섬웨어 탐지 및 분석 기술 개발)

  • Kang, Ho-Seok;Kim, Sung-Ryul
    • Journal of Digital Contents Society
    • /
    • v.19 no.2
    • /
    • pp.363-370
    • /
    • 2018
  • Ransomware detection has become a hot topic in computer security for protecting digital contents. Unfortunately, current signature-based and static detection models are often easily evadable by compress, and encryption. For overcoming the lack of these detection approach, we have proposed the dynamic ransomware detection system using data mining techniques such as RF, SVM, SL and NB algorithms. We monitor the actual behaviors of software to generate API calls flow graphs. Thereafter, data normalization and feature selection were applied to select informative features. We improved this analysis process. Finally, the data mining algorithms were used for building the detection model for judging whether the software is benign software or ransomware. We conduct our experiment using more suitable real ransomware samples. and it's results show that our proposed system can be more effective to improve the performance for ransomware detection.

Design of an Infant's App using AI for increasing Learning Effect (학습효과 증대를 위한 인공지능을 이용한 영유아 앱 설계)

  • Oh, Sun Jin
    • The Journal of the Convergence on Culture Technology
    • /
    • v.6 no.4
    • /
    • pp.733-738
    • /
    • 2020
  • It is really hard to find an infant's App, especially for the age under 5, even though there are lots of Apps developed and distributed nowadays. The selection of the proper infant's App is difficult since the infants' App should be useful, safe and helpful for the development of their intelligence. In this research, we design the useful infant's App for the development of their intelligence by applying the AI technology for increasing the learning effect in order to satisfy the characteristics of the infants' needs. A proposed App is the collection of interesting games for infants such as picture puzzle game, coloring shapes game, pasting stickers game, and fake mobile phone feature enables them to play interesting phone game. Furthermore, the proposed App is also designed to collect and analyze the log information generated while they are playing games, share and compare with other infants' log information to increase the learning effect. After then, it figures out and learns their game tendency, intelligibility, workmanship, and apply them to the next game in order to increase their interests and concentration of the game.

Multimodal Biometrics Recognition from Facial Video with Missing Modalities Using Deep Learning

  • Maity, Sayan;Abdel-Mottaleb, Mohamed;Asfour, Shihab S.
    • Journal of Information Processing Systems
    • /
    • v.16 no.1
    • /
    • pp.6-29
    • /
    • 2020
  • Biometrics identification using multiple modalities has attracted the attention of many researchers as it produces more robust and trustworthy results than single modality biometrics. In this paper, we present a novel multimodal recognition system that trains a deep learning network to automatically learn features after extracting multiple biometric modalities from a single data source, i.e., facial video clips. Utilizing different modalities, i.e., left ear, left profile face, frontal face, right profile face, and right ear, present in the facial video clips, we train supervised denoising auto-encoders to automatically extract robust and non-redundant features. The automatically learned features are then used to train modality specific sparse classifiers to perform the multimodal recognition. Moreover, the proposed technique has proven robust when some of the above modalities were missing during the testing. The proposed system has three main components that are responsible for detection, which consists of modality specific detectors to automatically detect images of different modalities present in facial video clips; feature selection, which uses supervised denoising sparse auto-encoders network to capture discriminative representations that are robust to the illumination and pose variations; and classification, which consists of a set of modality specific sparse representation classifiers for unimodal recognition, followed by score level fusion of the recognition results of the available modalities. Experiments conducted on the constrained facial video dataset (WVU) and the unconstrained facial video dataset (HONDA/UCSD), resulted in a 99.17% and 97.14% Rank-1 recognition rates, respectively. The multimodal recognition accuracy demonstrates the superiority and robustness of the proposed approach irrespective of the illumination, non-planar movement, and pose variations present in the video clips even in the situation of missing modalities.

A Study on Image Binarization using Intensity Information (밝기 정보를 이용한 영상 이진화에 관한 연구)

  • 김광백
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.8 no.3
    • /
    • pp.721-726
    • /
    • 2004
  • The image binarization is applied frequently as one part of the preprocessing phase for a variety of image processing techniques such as character recognition and image analysis, etc. The performance of binarization algorithms is determined by the selection of threshold value for binarization, and most of the previous binarization algorithms analyze the intensity distribution of the original images by using the histogram and determine the threshold value using the mean value of Intensity or the intensity value corresponding to the valley of the histogram. The previous algorithms could not get the proper threshold value in the case that doesn't show the bimodal characteristic in the intensity histogram or for the case that tries to separate the feature area from the original image. So, this paper proposed the novel algorithm for image binarization, which, first, segments the intensity range of grayscale images to several intervals and calculates mean value of intensity for each interval, and next, repeats the interval integration until getting the final threshold value. The interval integration of two neighborhood intervals calculates the ratio of the distances between mean value and adjacent boundary value of two intervals and determine as the threshold value of the new integrated interval the intensity value that divides the distance between mean values of two intervals according to the ratio. The experiment for performance evaluation of the proposed binarization algorithm showed that the proposed algorithm generates the more effective threshold value than the previous algorithms.

Detecting Salient Regions based on Bottom-up Human Visual Attention Characteristic (인간의 상향식 시각적 주의 특성에 바탕을 둔 현저한 영역 탐지)

  • 최경주;이일병
    • Journal of KIISE:Software and Applications
    • /
    • v.31 no.2
    • /
    • pp.189-202
    • /
    • 2004
  • In this paper, we propose a new salient region detection method in an image. The algorithm is based on the characteristics of human's bottom-up visual attention. Several features known to influence human visual attention like color, intensity and etc. are extracted from the each regions of an image. These features are then converted to importance values for each region using its local competition function and are combined to produce a saliency map, which represents the saliency at every location in the image by a scalar quantity, and guides the selection of attended locations, based on the spatial distribution of saliency region of the image in relation to its Perceptual importance. Results shown indicate that the calculated Saliency Maps correlate well with human perception of visually important regions.

Design of Variable Average Operation without the Divider for Various Image Sizes (다양한 영상크기에 적합한 나눗셈기를 사용하지 않은 가변적 평균기의 설계)

  • Yang, Jeong-Ju;Jeong, Hyo-Won;Lee, Sung-Mok;Choi, Won-Tae;Kang, Bong-Soon
    • Journal of the Institute of Convergence Signal Processing
    • /
    • v.10 no.4
    • /
    • pp.267-273
    • /
    • 2009
  • In this paper, we proposed a variable average operation for a WDR(Wide Dynamic Range). The previously proposed average operation [5] improves hardware efficiency and complexity by replacing divider with multiplier. However, the previously proposed method has some weak-points. For example, there are counting horizontal and vertical length, and then the multiplier selects a Mode set by the user when the lengths exactly correspond with the image's size in the Mode. To compensate some weak-points, we change a Mode selection methods as a using the image's total size. Also, we propose another feature that it can be applied to various image sizes. To get a more accurate average, we add an external compensation value. We design the variable average operation using a Verilog-HDL and confirm that the Serial Multiplier's structure is better efficiency than Split Multiplier's structure.

  • PDF

Deriving Local Association Rules by User Segmentation (사용자 구분에 의한 지역적 연관규칙의 유도)

  • Park, Se-Il;Lee, Soo-Wun
    • Journal of KIISE:Software and Applications
    • /
    • v.29 no.1_2
    • /
    • pp.53-64
    • /
    • 2002
  • Association rule discovery is a method that detects associative relationships between items or attributes in transactions. It is one of the most widely studied problems in data mining because it offers useful insight into the types of dependencies that exist in a data set. However, most studies on association rule discovery have the drawback that they can not discover association rules among user groups that have common characteristics. To solve this problem, we segment the set of users into user-subgroups by using feature selection and the user segmentation, thus local association rules in user-subgroup can be discovered. To evaluate that the local association rules are more appropriated than the global association rules in each user-subgroup, derived local association rules are compared with global association rules in terms of several evaluation measures.

Ensemble Classifier with Negatively Correlated Features for Cancer Classification (암 분류를 위한 음의 상관관계 특징을 이용한 앙상블 분류기)

  • 원홍희;조성배
    • Journal of KIISE:Software and Applications
    • /
    • v.30 no.12
    • /
    • pp.1124-1134
    • /
    • 2003
  • The development of microarray technology has supplied a large volume of data to many fields. In particular, it has been applied to prediction and diagnosis of cancer, so that it expectedly helps us to exactly predict and diagnose cancer. It is essential to efficiently analyze DNA microarray data because the amount of DNA microarray data is usually very large. Since accurate classification of cancer is very important issue for treatment of cancer, it is desirable to make a decision by combining the results of various expert classifiers rather than by depending on the result of only one classifier. Generally combining classifiers gives high performance and high confidence. In spite of many advantages of ensemble classifiers, ensemble with mutually error-correlated classifiers has a limit in the performance. In this paper, we propose the ensemble of neural network classifiers learned from negatively correlated features using three benchmark datasets to precisely classify cancer, and systematically evaluate the performances of the proposed method. Experimental results show that the ensemble classifier with negatively correlated features produces the best recognition rate on the three benchmark datasets.