결함 지역화는 관찰된 결함의 근본 원인을 자동 인식 하는 것이 가능하기 때문에 규모가 큰 분산시스템에서 중요 역할 수행하며 시스템의 신뢰성 개선을 위해 시스템의 관리와 제어가 가능한 자가 관리를 지원한다. 결함 지역화를 지원하는 기존 연구들은 유비쿼터스 환경에서 베이지안 네트워크와 같은 인공지능 기술들을 주로 사용하여 진단과 예측 기능 중 하나만을 고려하고 있다. 따라서, 본 논문에서는 시스템의 신뢰성 개선을 위해 실시간 시스템 성능 스트림에 대한 학습을 통해 자가관리를 위한 확률적 의존 분석을 기반으로 하는 결함 지역화 방법을 제안하여 진단과 예측기능을 동시 제공한다. 학습 방법으로 베이지안 네트워크 알고리즘을 사용하여 각종 관련된 요소들을 연결함으로써 네트워크를 생성하고 확률적 의존 관계를 통해 귀납적과 연역적 추론기능을 제공한다. 베이지안 네트워크의 구성은 노드들간의 연관성을 찾아내는 것이 중요하기 때문에 그것을 구성하는 인자의 개수가 많은 경우 노드 순서 리스트를 추출하는 사전처리 과정이 필요하다. 따라서 전체 모델링 프로세스에 대한 개선이 요구된다. 이러한 문제를 해결하기 위해 발생한 문제와 관련성이 높은 노드 순서 리스트를 추출하는 방법을 제공한다. 구조 학습을 지원 하는 사전처리 방법을 통해 다양한 문제 영역에서의 학습 효율성을 높이며 학습에 필요로 되는 시간을 줄인다. 제안 방법론을 통해서 시스템의 자원 문제를 신속하고 정확하게 진단하는 것이 가능하며, 관찰된 정보를 기반으로 실행 중에 발생되는 잠재적인 문제를 예측하는 것이 가능하다. 시스템 성능 평가 영역에서 제안 방법론을 적용한 시스템 성능 분석을 기반으로 진단, 예측의 효율성과 정확성을 평가하여 제안 방법론의 유효성을 입증하였다.
결함 지역화는 관찰된 결함의 근본 원인을 자동 인식 하는 것이 가능하기 때문에 규모가 큰 분산시스템에서 중요 역할 수행하며 시스템의 신뢰성 개선을 위해 시스템의 관리와 제어가 가능한 자가 관리를 지원한다. 결함 지역화를 지원하는 기존 연구들은 유비쿼터스 환경에서 베이지안 네트워크와 같은 인공지능 기술들을 주로 사용하여 진단과 예측 기능 중 하나만을 고려하고 있다. 따라서, 본 논문에서는 시스템의 신뢰성 개선을 위해 실시간 시스템 성능 스트림에 대한 학습을 통해 자가관리를 위한 확률적 의존 분석을 기반으로 하는 결함 지역화 방법을 제안하여 진단과 예측기능을 동시 제공한다. 학습 방법으로 베이지안 네트워크 알고리즘을 사용하여 각종 관련된 요소들을 연결함으로써 네트워크를 생성하고 확률적 의존 관계를 통해 귀납적과 연역적 추론기능을 제공한다. 베이지안 네트워크의 구성은 노드들간의 연관성을 찾아내는 것이 중요하기 때문에 그것을 구성하는 인자의 개수가 많은 경우 노드 순서 리스트를 추출하는 사전처리 과정이 필요하다. 따라서 전체 모델링 프로세스에 대한 개선이 요구된다. 이러한 문제를 해결하기 위해 발생한 문제와 관련성이 높은 노드 순서 리스트를 추출하는 방법을 제공한다. 구조 학습을 지원 하는 사전처리 방법을 통해 다양한 문제 영역에서의 학습 효율성을 높이며 학습에 필요로 되는 시간을 줄인다. 제안 방법론을 통해서 시스템의 자원 문제를 신속하고 정확하게 진단하는 것이 가능하며, 관찰된 정보를 기반으로 실행 중에 발생되는 잠재적인 문제를 예측하는 것이 가능하다. 시스템 성능 평가 영역에서 제안 방법론을 적용한 시스템 성능 분석을 기반으로 진단, 예측의 효율성과 정확성을 평가하여 제안 방법론의 유효성을 입증하였다.
Fault localization plays a significant role in enormous distributed system because it can identify root cause of observed faults automatically, supporting self-managing which remains an open topic in managing and controlling complex distributed systems to improve system reliability. Although many Ar...
Fault localization plays a significant role in enormous distributed system because it can identify root cause of observed faults automatically, supporting self-managing which remains an open topic in managing and controlling complex distributed systems to improve system reliability. Although many Artificial Intelligent techniques have been introduced in support of fault localization in recent research especially in increasing complex ubiquitous environment, the provided functions such as diagnosis and prediction are limited. In this paper, we propose fault localization for self-managing in performance evaluation in order to improve system reliability via learning and analyzing real-time streams of system performance events. We use probabilistic reasoning functions based on the basic Bayes' rule to provide effective mechanism for managing and evaluating system performance parameters automatically, and hence the system reliability is improved. Moreover, due to large number of considered factors in diverse and complex fault reasoning domains, we develop an efficient method which extracts relevant parameters having high relationships with observing problems and ranks them orderly. The selected node ordering lists will be used in network modeling, and hence improving learning efficiency. Using the approach enables us to diagnose the most probable causal factor with responsibility for the underlying performance problems and predict system situation to avoid potential abnormities via posting treatments or pretreatments respectively. The experimental application of system performance analysis by using the proposed approach and various estimations on efficiency and accuracy show that the availability of the proposed approach in performance evaluation domain is optimistic.
Fault localization plays a significant role in enormous distributed system because it can identify root cause of observed faults automatically, supporting self-managing which remains an open topic in managing and controlling complex distributed systems to improve system reliability. Although many Artificial Intelligent techniques have been introduced in support of fault localization in recent research especially in increasing complex ubiquitous environment, the provided functions such as diagnosis and prediction are limited. In this paper, we propose fault localization for self-managing in performance evaluation in order to improve system reliability via learning and analyzing real-time streams of system performance events. We use probabilistic reasoning functions based on the basic Bayes' rule to provide effective mechanism for managing and evaluating system performance parameters automatically, and hence the system reliability is improved. Moreover, due to large number of considered factors in diverse and complex fault reasoning domains, we develop an efficient method which extracts relevant parameters having high relationships with observing problems and ranks them orderly. The selected node ordering lists will be used in network modeling, and hence improving learning efficiency. Using the approach enables us to diagnose the most probable causal factor with responsibility for the underlying performance problems and predict system situation to avoid potential abnormities via posting treatments or pretreatments respectively. The experimental application of system performance analysis by using the proposed approach and various estimations on efficiency and accuracy show that the availability of the proposed approach in performance evaluation domain is optimistic.
* AI 자동 식별 결과로 적합하지 않은 문장이 있을 수 있으니, 이용에 유의하시기 바랍니다.
가설 설정
1) The larger the numbers of levels of considering components, the more generated rules are needed, which makes the system experience high overload and low efficiency.2) All rule or case generations should be user-defined in advanced. 3) All created rules or cases are impossible to be comprehensive, which implies that one event occurred may not be included in the existing aggregation.
Sixth, all parameters presented as pairs in the set are realigned to a single list.
제안 방법
For proving the effects of the proposed Bayesian network approach to fault localization for self-managing in performance evaluation, we apply testing data into the built model then compare the results with actual results. At first, we evaluate time consumption of structure learning and error rate given different numbers of parameters, showing that the obvious effect when using a certain number of parameters that are highly related with the domain.
In order to build Bayesian network structure for problem localization, it not only to find simple relationships between causes and effects but also to dig out the dependency relations between causes, which enable us to construct a more compact network for problem localization based on probabilistic inferences, as given in (Fig. 4).
Different from other existing researches on using Bayesian network, it adds preprocessing course to extract a certain parameters as a node ordering list for contributing to modeling Bayesian network efficiently. In order to prove the availability and efficiency of proposed approach, we perform it on system performance evaluation domain using the proposed fault localization for self-managing and make comparisons under different conditions.
Recently many methods for structure learning [14] are developed, finding the structure that is most probable to training data. In this paper, a certain parameters with order are taken as input to create a fine-grained model by analyzing conditional independency evaluation between all pairs of nodes. It should be emphasized that the Bayesian network implies conditional independencies via showing conditional probability tables for leaf nodes having direct parent nodes.
There are unsolved problems such as overfitting and generalization in recent works. In this paper, we propose an approach to fault localization based on Bayesian network learning to provide probabilistic dependency analysis which is used to localize or predict exact cause of performance problems under given observation in ubiquitous computing environment. We extract node ordering lists that derived from preprocessing course to construct probabilistic dependency model, which improves the efficiency of modeling without degrading the quality of learning.
In order to improve the performance of learning with domain knowledge, an improved learning process is provided before structure learning. Using the proposed method, we can create a hierarchical network that represents direct relationships between nodes with high efficiency and accuracy, which we use to make probabilistic dependency analysis to determine the exact root cause of system performance problems. Different from other existing researches on using Bayesian network, it adds preprocessing course to extract a certain parameters as a node ordering list for contributing to modeling Bayesian network efficiently.
이론/모형
In this paper, we use probabilistic machine learning method, which is mainly used as a modeling tool, to propose a dependency model structure for fault diagnosis and prognosis in self-managing systems. In terms of accuracy and efficiency of diagnosing problems and predicting potential problems, we can deal with the data in the raw beforehand, where the relative parameters in an order are extracted for the next modeling step.
성능/효과
However, recent fault localization techniques using machine learning approach such as rule-based or case-based inferences will bring problems because most of them rarely consider relationships between collected information, which are inefficient in the case of uncertainty.1) The larger the numbers of levels of considering components, the more generated rules are needed, which makes the system experience high overload and low efficiency. 2) All rule or case generations should be user-defined in advanced.
2) All rule or case generations should be user-defined in advanced.3) All created rules or cases are impossible to be comprehensive, which implies that one event occurred may not be included in the existing aggregation.
참고문헌 (17)
R. K. Sahoo, A. J. Oliner, I. Rish, M. Gupta, J. E. Moreira, S. Ma, R. Vilalta, and A. Sivasubramaniam, “Critical event prediction for proactive management in large-scale computer clusters,” In Proceedings of the ACM SIGKDD, Intl. Conf. on Knowledge Discovery and Data Mining, pp.426.435, August 2003
Jeffrey O. Kephart David M. Chess IBM Thomas J. Watson Research Center, “The Vision of Autonomic Computing,” IEEE Computer Society, January 2003
Irina Rish, Mark Brodie, Sheng Ma, Natalia Odintsova, Alina Beygelzimer, Genady Grabarnik, and Karina Hernandez, “Adaptive Diagnosis in Distributed Systems,” IEEE Transactions on Neural Networks, March 2005
Yuan-Shun Dai, “Autonomic Computing and Reliability Improvement,” Proceedings of Eighth IEEE International Symposium on Object-Oriented Real-Time Distributed Computing (ISORC'05), pp. 204-206, 2005
IBM Self-Aware Distributed Systems: http://domino. watson.ibm.com/comm/research.nsf/pages/r.ai.innovation.2. html
Sun Microsystems: Predictive Self-Healing in the Solaris 10 Operating System: http://www.sun.com/ bigadmin/content/selfheal 0
Bhaskara Reddy Moole and Raghu Babu Korrapati, “Enterprise web site problem diagnosis using Bayesian Belief Networks”, SoutheastCon, Proceedings, IEEE, pp. 384-396, 2005
J.Bronstein, A.Das., “Self-Aware Services- Using Bayesian Networks for Detecting Anomalies in Internet-based Services”, HP Labs Technical Reports HPL-2001-23R1, 2001
Rui Zhang, Steve Moyle and Steve McKeever, and Alan Bivens, “Performance Problem Localization in Self-Healing, Service-Oriented Systems using Bayesian Networks”, Proceedings of the 2007 ACM symposium on Applied computing, pp. 104-109, 2007
Malgorzata Steinder, Adarshpal S.Sethi, “Probabilistic Fault Localization in Communication Systems Using Belief Networks”, IEEE/ACM Transactions on Networking, pp.809-822, October 2004
Jianguo Ding, Bernd Kramer, Yingcai Bai, and hansheng Chen, “Backward inference in Bayesian networks for distributed systems management,” Journal of Network and Systems Management, Vol.13, No. 4, December 2005
Ethem Alpaydm, Introduction of Machine Learning. Massachusetts Institute of Technology, pp.39-60, 2004
Charles River Analytics Inc, About Bayesian Belief Networks, Charles River Analytics, Inc., 2004
Jie Cheng, David A. Bell,Weiru Liu, “An algorithm for Bayesian Belief Network construction from Data”, In Proceedings of AI &STAT', pp. 83-90, 1997
Cheng, J., Bell, D. and W. Liu, “Learning Bayesian Networks from Data: An Efficient Approach Based on Information Theory”, In Proceedings of the sixth ACM International Conference on Information and Knowledge Management, 1997
※ AI-Helper는 부적절한 답변을 할 수 있습니다.