The problem of estimating the performance of a given classifier on a given data set is discussed for the case when no knowledge is available concerning the underlying distributions. A new method of estimating the prob...
详细信息
The problem of estimating the performance of a given classifier on a given data set is discussed for the case when no knowledge is available concerning the underlying distributions. A new method of estimating the probability of misclassification is proposed which yields essentially unbiased results similar to Lachenbruch's U-method with far less computation involved. While no theoretical work is presented, a practical rule of thumb is given for choosing the parameters of the estimator. The results are based on experiments performed with a data set concerning six diseases related to epigastric pain, and underline the importance of reporting performance on both the testing data and the training data. Whereas previous papers have continually reported results with a probability of correct classification as high as 74. 3 per cent on the raw data and 92. 0 per cent on ″processed″ data, in this paper it is shown that a much more significant estimate of the probability of correct classification based on this data set is 51. 0 per cent.
Colloidal silver particles are formed on float glass by heat treatment with coated silica film containing Ag by sol-gel process . The Sn2+ on surface of float glass influences the formation of colloidal Ag particles ....
详细信息
Colloidal silver particles are formed on float glass by heat treatment with coated silica film containing Ag by sol-gel process . The Sn2+ on surface of float glass influences the formation of colloidal Ag particles . The microstructure of the film and the granularity of silver particles were studied by TEM and HEED. The emission property of the samples was measured. The results show that aggregation of metal particles degrades emission intensity , and that content of Ag, withdrawing speed and heat treatment temperature of samples has a greater effect on photoluminescence.
Silicon nitride composite is joined to itself by heating interlayer of Y2 O3 -AL2O3 -SiO2 mixtures above their liquidus temperatures in flowing nitrogen. The joined specimens are tested in four point flexure from room...
详细信息
Silicon nitride composite is joined to itself by heating interlayer of Y2 O3 -AL2O3 -SiO2 mixtures above their liquidus temperatures in flowing nitrogen. The joined specimens are tested in four point flexure from room temperature to 1373 K. The interface microstruclure and fractured surfaces after testing are observed and analyzed by SEM, EPMA and XRD respectively. The results show that F2 O3 -A12 O3 -SiO2 glass reacts with Si3 N4 at interface, forming the Si3 N4/Si2 N2 O( Y-AlrSi-O-N glass/ Y-Al- Si-O glass gradient interface. With the increase of bonding temperature and holding time, the joint strength first increases, reaching a peak, and then decreases . According to interfacial analyses , the bonding strength depends on joint thickness .
Feature extraction is a key issue to machine condition monitoring and fault diagnosis. The features must contain the necessary discriminative information for the fault classifier to have any chance of accurate classif...
详细信息
Feature extraction is a key issue to machine condition monitoring and fault diagnosis. The features must contain the necessary discriminative information for the fault classifier to have any chance of accurate classification. This paper presents a study that uses principal component analysis to reduce dimensionality of the feature space and to get an optimal subspace for machine fault classification. Industrial gearbox vibration signals measured from different operating conditions are analyzed using the above method. The experimental results indicate that the method extracts diagnostic information effectively for gear fault classification and has a good potential for application in practice.
Dispatching rule-based sch.duling is a kind of dynamic sch.duling commonly used in real world applications. Because of the lack of sch.duling objective, it cannot optimize the specific performances at which shop manag...
详细信息
Dispatching rule-based sch.duling is a kind of dynamic sch.duling commonly used in real world applications. Because of the lack of sch.duling objective, it cannot optimize the specific performances at which shop managers aim in the current production period. To overcome the limitations of the dispatching rule-based sch.duling, an iterative learning sch.duling sch.me is proposed in this paper. A sch.duling objective function, which reflects the performance criteria in which the shop managers are most interested, is established and used to guide the optimization of the crucial performances. According to the value of the sch.duling objective obtained from the last simulation period, the parameters are adjusted so as to decrease the objective during the next simulation period. Experimental results show that the iterative learning sch.duling overcomes the limitations of the dispatching rule-based sch.duling and achieves higher performances.
The competitive learning technique is a well-known algorithm used in neural networks, which classifies the input vectors, so that the vectors (samples) belonging to the same class have similar characteristics. Dynamic...
详细信息
ISBN:
(纸本)0780375084
The competitive learning technique is a well-known algorithm used in neural networks, which classifies the input vectors, so that the vectors (samples) belonging to the same class have similar characteristics. Dynamic competitive learning is an unsupervised learning technique, which consists of two additional parts related to conventional competitive learning: a method of generation of new units within a cluster and a method of generating new clusters. The model is capable for the high-level storage of complex data structures, whose classification include exception handling.
CURE (Clustering Using Representatives) is an efficient clustering algorithm for large databases, compared with other clustering methods, which is more robust to outliers, and identifies clusters having non-spherical ...
详细信息
ISBN:
(纸本)0780375084
CURE (Clustering Using Representatives) is an efficient clustering algorithm for large databases, compared with other clustering methods, which is more robust to outliers, and identifies clusters having non-spherical shapes and wide variances in size. CURE employs a fixed number of representative points to describe the cluster, and the set of representative points are firstly chosen randomly, and then are shrunk toward the mean of cluster. The shrinking operation plays a key role in CURE, which is used for weakening the effect of outliers. But we find that the shrinking sch.me of CURE is dependent on a hidden assumption of spherical shape of cluster, therefore CURE has difficulties in dealing with databases having specific shapes. In this paper, CURE-NS (CURE with New Shrinking sch.me) is proposed to overcome this problem, which uses the difference of density values of the representative points to determine the direction and distance of shrinking. Our shrinking sch.me has nothing to do with the shape of cluster. A range of experiments has demonstrated that CURE-NS has better clustering performance than CURE.
Magnesium phosphate cement-based binder(MPB)for repair of concretewas prepared by prop- ortionally mixing over burned MgO powder(M)withNHH_2PO_4 powder(P)and set modifying admixtures. It is characteristicby excellent ...
详细信息
Magnesium phosphate cement-based binder(MPB)for repair of concretewas prepared by prop- ortionally mixing over burned MgO powder(M)withNHH_2PO_4 powder(P)and set modifying admixtures. It is characteristicby excellent properties such as rapid setting, high strength and highbond strength to old con- crete. The study is focused on the keyfactors influencing the setting time and strength of MPB, the bondproperty of MPB to old concrete and the kinetic feature of thehydration of MPB.
This paper describes a smart and efficient approach of information extraction for Chinese free text. Automatic learning algorithm of pattern rule for Chinese free text and employment of heuristic information are depic...
详细信息
ISBN:
(纸本)0780375084
This paper describes a smart and efficient approach of information extraction for Chinese free text. Automatic learning algorithm of pattern rule for Chinese free text and employment of heuristic information are depicted. The methods that combine pattern rule matching with heuristic information is utilized to perform Information Extraction task. Experiment proves this method to be effective on improving extraction result of complex situations and increasing recall noticeably.
POST (part of speech) tagging is the basic work in Natural Language Processing (NLP). The tagging precision will have an important effect on the result of later process, such as syntax analysis. In this paper, a Chine...
详细信息
ISBN:
(纸本)0780375084
POST (part of speech) tagging is the basic work in Natural Language Processing (NLP). The tagging precision will have an important effect on the result of later process, such as syntax analysis. In this paper, a Chinese POS tagger based on maximum entropy model is presented, which trains from a large corpus annotated with Chinese POS tags and assigns a best tag sequence to a Chinese sentence to be annotated. In this model, all the features that are useful to predicate the POS tags are mined to make the model close to the real case. In addition, for the problem of overfitting, a smoothing method and a POS dictionary are maintained to reduce the model's dependence to training data and improve the efficiency of searching process. Open testing results shows that Chinese POS tagging with this method can achieve the state-of-art accuracy (96.8%).
暂无评论