Search results

Automatic text classification is a crucial natural language processing (NLP) technique for handling and organizing text data.
The feature selection methods can effectively remove the noise in the feature space, but are lack of ability to tackle with data sparseness.
Experiments and results Datasets and data preprocessing.The Chinese text classification data sets of Sougou news corpus, which are multi-classification text sets, are adopted in our experiments.
Unfortunately the dimension of text classification data sets is usually very high.
Knowl.Data Eng. 21(3): 428-441 [6] Li Zhixing, Xiong Zhongyang, Zhang Yufang, Liu Chunyong, Li Kuan.

It is possible to add data sent directly from the vehicle diagnostic system to the central unit of the analyzer and use the GPS signal (Global Positioning System).
Data storage capacity Over 10 hours at data acquisition rate of 1 Hz 3.
In this time the system recorded (with the resolution of 1 second) the concentrations of the exhaust components, the exhaust mass flow and data from the on-board diagnostic system of the excavator.
In the case of the data acquired from the machine OBD the data related to the engine parameters were of particular importance i.e. engine speed and engine load.
NOx and PM are key components for diesel engines and their reduction still poses a serious problem despite the accessibility to advanced technical solutions such as PM filters and SCR systems.

Data sources.
The RSSI data of this paper was gotten by adding different kinds of noise after choosing model to calculate the RSSI data.
Choosing different propagation models to calculate the RSSI values of the point to be measured and increasing the noise of different channel to get the simulative input data.
We make the RSSI simulation values of test points as data source, build the corresponding propagation model in different regions and use Gaussian filter to handle the obtained data for removing singular value.
HORUS - high-dimensional Model Order Reduction via low moment-matching upgraded sampling.

A 50 per cent reduction in these additives would indicate the need for an oil change.
DISCUSSION AND CONCLUSIONS Input of test data into a dedicated data processor is now the norm in modern analysis techniques, regardless of whether the origin of the test data is from laboratory or field equipment.
The physical test data can be measured against accepted standards of used oil control and condemnation limits recognised by learned societies and oil companies (IP, ASTM).
Based on successive test data, predictive trends are established in each test category thus effectively eliminating unnecessary oil changes.
The accumulation of wear debris test data over a predetermined sample period establishes a normality of wear generation pattern either in elemental concentration quantification as in spectrometry, or in particle size quantification as in ferrography, the latter being more practical in many cases.

According to current literature data cancer of larynx and laryngopharynx accounts for 5-6% of all malignancies, moreover, this localization takes 60-70% of all malignant tumors of the upper respiratory tract.
Traditionally, the evaluation of the effectiveness of neoadjuvant chemotherapy (NACT) is conducted according to the data of clinical examination, fiberoptic laryngoscopy with withdrawal of biopsy material for the study of the degree of drug pathomorphism and data of ultrasound of regional lymph nodes.
Data obtained by polypositional scanning were processed using specialized processing software package E.Soft by company "Siemens" (Germany).
In our case, complete tumor regression after NACT was noted in a single observation, and we believe this event could not have a significant impact on the importance of statistic data.
Nevertheless, the results suggest that the decline in the indices T/SSGmax and T/PGmax and T/STEAmax during chemotherapy indicates its efficiency, which in turn is supported by clinical data and the results of morphological studies.

The asset management (AM) system of a transformer station (ST) Stages of system implementation An A.M. system seeks to quantify the electrical equipments state on the base of historical and real time data and permits to establish a hierarchy of maintenance activities [2].
The diagnosis system selects data from the monitoring systems and compares the measured values ​​with thereshold values​​.
The second section (Fig. 6) performed on-line, monitoring the equipment, data acquisition and processing, gives information about the technical status of the monitored equipment.
The window diagram (Fig. 8) contains software for real-time monitoring, data aquisiton and their processing in order to quantify the technical condition of the equipment in the group.
Fig. 8 Single line diagram of transformer station Fig. 9 FI calculation For determining priority of the maintenance activity and to check the relevance of the data provided by the implemented model considered two cases: Case I: 8 monitored equipment by 5 parameters surveillance / equipment.

By using a vibration signal collective and analytical system that is aimed at a Φ3×11m edgetransmission ball mill system and is according to the existing violent vibration phenomenon of the machine system, we execute the field test to acquire the related data and process the analysis, and then we accurately find the main fault reasons and the validly accessible solutions.
Specification and Data Acquisition Table 1 shows the specification of the ball mill system, it found that the foundation of the ball mill system vibrated with great abnormal magnitude; as well, the foundation of the main decelerator and the main electromotor had the same severe phenomenon.
The measured points can provide valid vibration data for the fault detection and diagnosis.
To be convenient for comparing the faulty data with the normal data, the points 2# and 3# that are located on the other ball mill system of Φ3.8×13m with 2500kw output power are selected, and the points 2# and 3# are relative to the same positions as points 2 and 3 of the faulty system.
In addition, the data for the normal ball mill system is acquired to provide useful evidence for the fault diagnosis.

Equally important is the choice of the method of processing of raw data of sedimentation analysis, ie choice of the method of processing of the sedimentation curve [3].
The data of sedimentation analysis are in good agreement with the results of measurements of the dependence of electrokinetic potential of particles of coal powder in an aqueous medium on the duration of grinding.
Further reduction of z-potential may be due to the consolidation of fine powder particles of coal, which is confirmed by the differential distribution curves of coal particles by size.
In addition, data from the IR spectra can provide information about the link between different structural groups.
These data suggest that during the mechano-chemical processing of coal the profound degradation of coal macromolecules occurs, which affects the yield of distillate of coal.

Hardness curve of laser remelting coating presented three gradients, namely high hardness area, hardness reduction area and low hardness area.
It was evident that the fracture mode of plasma spraying samples was the fracture between coating and substrate, so the experimental data shown in Table 1 were bond strength of coating and substrate.
So data shown in Table 2 were not the bond strength of laser remelting coating and substrate, it was just the bond strength of adhesive.
But through the observation of coating fracture mode and the comparison analysis data of experiment data, we could draw the conclusion that bond strength of coatings after laser remelting treatment was far greater than the bond strength of adhesive as well as the bond strength of plasma spraying coating.
Table 1 Tensile testing data of plasma spraying coating N1-1 N2-2 N3-3 N4-4 N5-5 AVG Maximum fracture load [KN] 7.653 7.546 7.725 7.664 7.708 7.659 Bond strength [Mpa] 24.3 24.0 24.6 24.4 24.5 24.36 Table 2 Tensile testing data of laser re-melting coating N1-1 N2-2 N3-3 N4-4 N5-5 AVG Maximum fracture load [KN] 20.9 21.2 20.5 21.6 21.0 21.04 Bond strength [Mpa] 66.6 67.5 65.3 68.8 66.9 67.02 Analysis of Wear Resistance Test.

According to this algorithm, if is the discrete sampling data of original signals , then the decomposition formula of orthogonal wavelet transform of the signal is: (5) In the formula, is the approximate coefficient; is the detail coefficient; and are respectively low-pass filter and high-pass filter; is the number of decomposition layer; is number of the discrete sampling point.
Data Description Sample data come from the bearing data center of Case Western Reserve University [6].
In order to obtain the characteristic frequency, the relevant program is developed based on MATLAB platform, and relevant analysis on fault data is completed by using discrete wavelet transform.
Wavelet Analysis Data Processing Using the discrete wavelet transform, the vibration acceleration signals of rolling bearing are converted to time-scale domain, then, the power spectrum analysis of the high frequency detail signal is completed.
After that, such processing are completed, including wavelet decomposition, noise reduction, reconstruction, Hilbert transform, etc, the characteristics frequency of rolling bearing is accurately found from its power spectrum, which shows the accuracy and effectiveness of this method.