Location prediction is a special case of spatial datamining classification. For instance, in the public safety domain, it may be interesting to predict location(s) of crime hot spots. In this study, we present Suppor...
详细信息
ISBN:
(纸本)1424402115
Location prediction is a special case of spatial datamining classification. For instance, in the public safety domain, it may be interesting to predict location(s) of crime hot spots. In this study, we present Support Vector machine (SVM) based approach to predict the location as alternative to existing modeling approaches. SVM forms the new generation of machinelearning techniques used to find optimal separability between classes within datasets. Experiments on two different spatial datasets show that SVMs gives reasonable results.
In the last few years neural network is found as an effective tool for patternrecognition. the success rate for recognizing known and unknown pattern is relatively, very high with compare to other techniques. this pa...
详细信息
ISBN:
(纸本)9789843238146
In the last few years neural network is found as an effective tool for patternrecognition. the success rate for recognizing known and unknown pattern is relatively, very high with compare to other techniques. this paper presents a comparative study of how neural network classifies the patterns from training data and recognizes if testing data holds that patterns. For learning from the training data lots of approaches are present among which we have selected the back-propagation method. Back-propagation algorithm in a feed-forward network. is used for the feature extraction. We have used two approaches and network was trained with specified data. We intended to find the match ratio of training pattern to testing pattern and the result data set found from the experiment also given in the paper.
the proceedings contain 115 papers. the topics discussed include: strategy coordination approach for safe learning about novel filtering strategies in multi agent framework;modeling and design of agent based open deci...
详细信息
ISBN:
(纸本)3540335846
the proceedings contain 115 papers. the topics discussed include: strategy coordination approach for safe learning about novel filtering strategies in multi agent framework;modeling and design of agent based open decision support systems;particle filter method for a centralized multisensor system;design and analysis of a novel load-balancing model based on mobile agent;construction and simulation of the movable propeller turbine neural network model;research and application of datamining in power plant process control and optimization;an efficient algorithm for incremental mining of sequential patterns;repeating pattern discovery from audio stream;a method to eliminate incompatible knowledge and equivalence knowledge;a similarity-aware multiagent-based web content management;evolutionary multi-objective optimization algorithm with preference for mechanical design;and refinement of fuzzy production rules by using a fuzzy-neural approach.
the increasing amount of data used for classification, as well as the demand for complex models with a large number of well tuned parameters, naturally lead to the search for efficient approaches making use of massive...
详细信息
ISBN:
(纸本)3540462910
the increasing amount of data used for classification, as well as the demand for complex models with a large number of well tuned parameters, naturally lead to the search for efficient approaches making use of massively parallel systems. We describe the parallelization of support vector machinelearning for shared memory systems. the support vector machine is a powerful and reliable datamining method. Our learning algorithm relies on a decomposition scheme, which in turn uses a special variable projection method, for solving the quadratic program associated with support vector machinelearning. By using hybrid parallel programming, our parallelization approach can be combined withthe parallelism of a distributed cross validation routine and parallel parameter optimization methods.
Sensors have been used with various purposes in the human life. A sensor which can be functioned as a part of a signal process unit or a mechanical machine is defined as "a part of a measuring instrument which de...
详细信息
ISBN:
(纸本)9789806560840
Sensors have been used with various purposes in the human life. A sensor which can be functioned as a part of a signal process unit or a mechanical machine is defined as "a part of a measuring instrument which detects and responds immediately changes of a environment". As a sensor just reports the voltage level respect to detected physical or chemical quantity, it is needed to convert properly into meaning data. In most of cases, a sensor array, which consists of various kinds of sensors are used to detect a environment. there are two classes of methods to analyze signal patterns from a sensor array;the statistical method and the neural network method. One method has weak points comparing with another. One of each method's weak points is that most of statistical methods cannot consider shape characteristics of the signal pattern and neural network methods take too long time in the learning process. In spite of this weakness, the neural network process has been used in most of gas patternrecognition in recent studies. In this paper, we introduce a statistical method using state transition model for gas recognition. this paper focuses on making the accurate state transition model. We call this state transition model as ADSTM(Angle Difference based State Transition Model). through various experiments, we analyze the proposed ADSTM modeling method. the results of experiments show that ADSTM is a fast and reliable statistical method for recognizing a signal pattern of the sensor array.
As more and more real-time data is sent to databases by DAS, large amounts of data are accumulated in power plants. Abundant knowledge exists in historical data but it is hard to find and summarize this in a tradition...
详细信息
ISBN:
(纸本)3540335846
As more and more real-time data is sent to databases by DAS, large amounts of data are accumulated in power plants. Abundant knowledge exists in historical data but it is hard to find and summarize this in a traditional way. this paper proposes a method of operation optimization based on datamining in a power plant. ne basic structure of the operation optimization based on datamining is established and the improved fuzzy association rule mining is introduced to find the optimization values from the quantitative data in a power plant. Based on the historical data of a 300MW unit, the optimal values of the operating parameters are found by using datamining techniques. the optimal values are provided to guide the operation online and experiment results show that excellent performance is achieved in the power plant.
To diagnose a slight and incipient fault in a power plant thermal system correctly and timely, a new fault recognition approach is put forward by using fault symptom zoom technology(SZT) and fuzzy patternrecognition ...
详细信息
ISBN:
(纸本)3540335846
To diagnose a slight and incipient fault in a power plant thermal system correctly and timely, a new fault recognition approach is put forward by using fault symptom zoom technology(SZT) and fuzzy patternrecognition method. By studying the rules of the faults pertinent to energy and mass balance in a power plant thermal system, a new fault symptom preprocessing method, which is called "fault symptom zoom technology", is put forward to preprocess the fault characteristic parameters. the complexity of the thermal system fault knowledge library can be effectively reduced and the slight fault recognition ability can be. greatly enhanced with SZT. the fault fuzzy patternrecognition method is introduced. A new general-purpose fuzzy recognition function is given, which can fit for various kinds of fault symptoms and is with favorable fault classifying ability. Some examples for incipient and slight fault diagnosis for a power plant thermal system are given to verify the effectiveness of the method.
We investigate the use of independent component analysis (ICA) for speech feature extraction in digits speech recognition systems. We observe that this may be true for recognition tasks based on Geometrical learning w...
详细信息
ISBN:
(纸本)3540335846
We investigate the use of independent component analysis (ICA) for speech feature extraction in digits speech recognition systems. We observe that this may be true for recognition tasks based on Geometrical learning with little training data. In contrast to image processing, phase information is not essential for digits speech recognition. We therefore propose a new scheme that shows how the phase sensitivity can be removed by using an analytical description of the ICA-adapted basis functions. Furthermore, since the basis functions are not shift invariant, we extend the method to include a frequency-based ICA stage that removes redundant time shift information. the digits speech recognition results show promising accuracy. Experiments show that the method based on ICA and Geometrical learning outperforms HMM in a different number of training samples.
this paper presents a novel unsupervised fuzzy clustering method based on clonal selection algorithm for anomaly intrusion detection in order to solve the problem of fuzzy k-means algorithm which is particularly sensi...
详细信息
ISBN:
(纸本)3540335846
this paper presents a novel unsupervised fuzzy clustering method based on clonal selection algorithm for anomaly intrusion detection in order to solve the problem of fuzzy k-means algorithm which is particularly sensitive to initialization and fall easily into local optimization. this method can quickly obtain the global optimal clustering with a clonal operator which combines evolutionary search, global search, stochastic search and local search, then detect abnormal network behavioral patterns with a fuzzy detection algorithm. Simulation results on the data set KDD CUP99 show that this method can efficiently detect unknown intrusions with lower false positive rate and higher detection rate.
this paper presents a novel approach for adaptive online multi-stroke sketch recognition based on Hidden Markov Model (HMM). the method views the drawing sketch as the result of a stochastic process that is governed b...
详细信息
ISBN:
(纸本)3540335846
this paper presents a novel approach for adaptive online multi-stroke sketch recognition based on Hidden Markov Model (HMM). the method views the drawing sketch as the result of a stochastic process that is governed by a hidden stochastic model and identified according to its probability of generating the output. To capture a user's drawing habits, a composite feature combining both geometric and dynamic characteristics of sketching is defined for sketch representation. To implement the stochastic process of online multi-stroke sketch recognition, multi-stroke sketching is modeled as an HMM chain while the strokes are mapped as different HMM states. To fit the requirement of adaptive online sketch recognition, a variable state-number determining method for HMM is also proposed. the experiments prove boththe effectiveness and efficiency of the proposed method.
暂无评论