Search results

Gabor facial feature selection using Adaboost For face image feature, if we use five scales and eight directions in the two-dimensional Gabor wavelet transform, it can produces 40 × M × N dimension of high dimensional feature vectors for an M × N samples, so we must conduct dimensionality reduction.
Using PCA (principal component analysis) to dimensionality reduction, we need matrix operations and consume a certain amount of computing time and storage space, and if the dimensions reduce excessive, then it cannot fully reflect the differences between the samples.
In order to solve this problem, a more direct method is to select a small part from the high dimensional feature that has a strong ability to distinguish to form the classifier. thus avoiding the dimensionality reduction operation performed in the classification process.
Figure 3 Level detection classifier Negative sample 1 Negative sample 2 Negative sample 3 Positive sample Strong classifier 1 Strong classifier 2 Strong classifier 3 The experiment results and analysis Do the experiment using YaLe face database.The data is made up of 15 people, 11 images of each, and the resolution is 320 x 243 pixels.
In order to test the effect of feature selection with Adaboost and the classifier performance, table 2 compared with the recognition effect using different dimension reduction methods and different classifier.

Because it is difficult to have quantitative data, we usually think that the bigger is the compression ratio, the better is the organization and performance.
In the process of Fig.3 (b) - Fig.3 (e), with the increase of reduction, the amount of metal width increases too.
The reduction at segmentation place is greater than other places and equivalent stress also is such.
In Fig.3 (e), mainly because after replacing the hole pattern ,the rolling reduction reduced, and the lateral width spread is no longer limited by the roll ring, the stress of each position has declined, so that metal go along the leg upward mobility.
As can be seen from Fig.3 (f) BD2 cogging mill first pass ~ Fig.3 (i) BD2 cogging mill fourth pass, because the rolling reduction of Fig.3 (f) and Fig.3 (h) mainly concentrate in web plate, equivalent stress is also mainly concentrated in the web plate, respectively 214Mpa and 253Mpa.The equivalent stress of the junction of web plate and flange is relatively small, respectively 178Mpa and 191Mpa.

Fig. 1 The four-vortex wake system constructed by the test model Fig. 2 The test model in the water towing tank The research was carried out in a water towing tank which equipped with Particle Image Velocimetry (PIV) measuring system which was employed in measuring and analyzing the experimental data.
wingspan 1 wingspan 2 wingspan 3 wingspan 4 wingspan 5 wingspan 6 wingspan 7 wingspan 8 wingspan 9 wingspan 10 wingspan 11 wingspan 15 wingspan 20 wingspan 25 wingspan 30 wingspan 35 wingspan 40 wingspan 43 Fig. 4 Evolution of the wake vortex vorticity of modified mode As soon as the generation of the vortex pair, the secondary vortex rotated around the primary vortex anticlockwise and then was absorbed by the primary vortex at about wingspan 5, which activated the Rayleigh-Ludwieg instability of the wake system, leading to the fracture of the primary vortex core and the reduction of vorticity indicated by the scalar map.
The efficiency of this study was judged by the percentage of circulation reduction η which was calculated by , (4) where Γmin was the statistical average circulation at last five wingspans and Γmax was the five maximum circulation.
According to Eq. 3 and Eq. 4, the circulation reduction of the modified model was 39.6% whereas 13.1% of the clean model in 43 wingspans.
The dissipation of wake vortex can only be accelerated instead of totally diminished. 4) The circulation of the wake flow field continually increases in about 10 wingspans and reaches to its climax between 5 to 15 wingspans. 5) Under the study of up to 43 wingspans, the circulation reduction of modified model is 39.6% whereas 13.1% of the clean model, proving the effectiveness of the tailored-flap treatment in eliminating the wake vortex under labor conditions.

The reduction of the CoO and FeO content is accompanied by a overall reduction of their grain sizes, from 21.3 nm at Tcal =400℃ to 8.4 nm at Tcal =800℃.
This is a consequence of the reduction of the amount of CoO and FeO phases, which reduces their antiferromagnetic contribution to the hysteresis loops, with a corresponding dominance of the magnetic contribution of the CoFe2O4 phase.
The presence of CoO and FeO caused a shift of the magnetic hysteresis cycles and a reduction of their coercivity and remanence ratio as compared to the bulk CoFe2O4.
New York:Wiley(1994) [11] Powder Diffraction File, Joint Committee on Powder Diffraction Standards, International Centre for Diffraction Data.

All the IR data confirmed the presence of the nitro group (C-NO2), and the 1H and 13C NMR analyses indicated that the nitration occurred exclusively at the C-3 position on hydroxyquinone.
Mechanism of synthesis of 2-hydroxy-3-nitro-1,4-naphthoquinone (2) The selective reduction of aryl nitro compounds in the presence of carbonyl, generally, stannous chloride in concentrated hydrochloric acid is one of the classic reagents for reduction of nitro to the target amine, because stannous chloride fails to reduce carbonyl group under the condition of reduction of nitro.
However, in our present study reduction with stannous chloride led to the formation of transitional product of 3-amino-l,2,4-trihydroxynaphthalene since both carbonyl and nitro groups were reduced.

These ceramics devices have replaced metallic resonant cavities successfully with significant size reduction, by a factor of the order of εr 1/2, favoring the miniaturization of microwave devices such as oscillators and filters [1].
Table 2 - Data of grain size deduced by Scherrer equation [nm] Milling time (h) * Pico TiO2 ♦Pico ZrO2 ♦Pico ZrO2 +Pico ZTS 5 66,1 36,6 29,4 40,8 15 45,9 44,3 23,9 17,7 30 26,6 29,8 26,3 --------- 45 --------- 17,0 15,6 --------- 60 --------- 9,7 14,7 --------- In general, there was a markedly decrease in grain size and it can be said that ball milling was very efficient in grain refinement.
It has been suggested that mechanical activated synthesis occurs in two stages: in the first one, there is a grain size reduction followed by a second stage, when the reaction takes place [11,12].
In this case both, reaction and grain size reduction, took place simultaneously.
When grain size reduction occurs, it is expected a strong agglomeration of the resuting powder.

In addition, the lagging quality of operating personnel and management are led to a lack of historical data.
Consider of randomicity and continuity of hydro-meteorological sequence, it takes the moving average for original data to weaken the influence of outliers[4,5].
(23) After an accumulative generation again, formatting the sequence: (24) Also, estimated parameters,: (25) Obtained estimates of,represented ,: (26) The original data sequence model sets up[8]: (27) Instance analysis Selecting monthly output of a small hydropower station from 2009 to 2011 as historical data predicts monthly data of 2012.
The original data is shown in Tab.1.
In addition, it is easy to collect data about hydrology and meteorological.

Since data traffic model is non-stationary and chaos with complex data structure, SVM will provide good traffic prediction for actual network data transmission.
Secondly, the prediction module forecasts the data packets and DRX receiver starts to receive all the data.
Any data from data source will make the program enter receiving module.
The principles are same and any detection of data during on duration time will change the program to receive data.
Thirdly, before receiving current data, it will check the sequence of current data.

The spectrums after preprocessing also contain a large amount of data.
The principal component analysis reduced the data dimension and made it easier to do the further analysis.
Hence, the two principal components did not loss the key information of the spectrums and the spectral analysis was simplified for the reduction of the amount of the data.
After preprocessing of the spectral data, the principal component analysis was applied to reduce the dimensions of the data, and obtained the principal component value, the Eigen value, the variance contribution and the accumulative variance contribution.
The principal component analysis reduced the data dimension and made it easier to do the further analysis.

Vibration and acoustic data were collected simultaneously during braking to identify the braking condition.
The data analysis focuses on the low frequency domain.
The determination of the source will aid in brake noise reduction efforts and reinforce the vibration analysis method as a system identification method for brake noise.
THE DATA IS ANALYZED IN THE FORM OF SOUND SPECTRUM AND VIBRATION SPECTRUM.