Search results

Both image and audio data are selected for experiments and comparisons.
However, most of existing multimedia retrieval systems only retrieve multimedia database which contains single mode data, or are able to deal with media data of multiple modes, but do not support cross-media retrieval.
Heterogeneous issue of different modal data characteristics is solved through subspace map.
There were 100 images and 70 audio data in each semantic category respectively, among which 60 images and 60 audio examples as training data, and the remaining 400 images and 100 audio data as "new" media object.
This method is based on analyzing the lower content feature of different modal media data, therefore, in addition to the audio and image data in the experiment, also applied to the other two kinds of multimedia objects.

Multi-dimensional data indexing has received much attention in a centralized database.
These methods divide the data space into smaller subspaces (regions).
CAN can be used to directly index multi-dimensional data in its natural space.
Most CAN systems such as use space filling curves to map multi-dimensional data to one dimensional data.
It is based on k-d-tree [3] to partition the data space, and then maps the data space into skip graph overlay network by encoding it into a unique key.

In the proposed system, all subsystems repeat their own optimal decision making as well as exchange production data with others.
The problem is to determine the material requirement planning, distribution planning and a logistics planning, so as to to make effort of minimizing total costs in terms of reduction in chain-wide inventory.
From the designed model, information from MPs to LCs is static data, including date related to batch, lead time, duedate, etc.; while the LCs would provides the dynamic real-time information to the MPs, including material requirements,production cycle, processing conditions and so on.
Information related to material requirement and other static data was send to the workshop by MPs, and then vehicles start to transport materials according to planned routes and expected time.
All subsystems repeat their own optimal decisions making as well as data exchanging with others, that made the whole system a closed-loop system for material and logistics optimization.

Experiment Data and Analysis 7-level wavelet de-noising When film is moving and C-frame structure is being static, double coating film thickness distribution data and level 1-7 wavelet de-noising data in industrial production are described in table 2.
Data of Fixed Threshold Value Suppose threshold value scale is 170μm-200μm, diagram 4 illustrates the original data of static C-frame structure and moving film as well as moving C-frame structure and moving film, data processed by level-3 wavelet, threshold output data and output data processed by level-7 wavelet.
Table 1 shows data analysis of each group.
Measured Data of Changed Threshold Value When threshold value scale is changed, diagram 4 shows original data and output data of different threshold values of static C-frame structure and moving film as well as moving C-frame structure and moving film.
Consequently, output data within film thickness scale 180μm -190μm decreases.

In the following, the experimental data is compared with the calculation of the model.
It demonstrates that in two approaches, the temperature appears at the maximum value at about 10 seconds, and then declines quickly during the forming stage due to the reduction of the resistance.
From the figures presented above, it can be inferred that the reduction of the temperature is the main reason for the increasing forming force.
As a result, the reduction of the cooling rate by increasing the current intensity during the forming is a feasible method to optimize the forming process.
During the forming process, the forming force increased with the reduction of the temperature.

Pt circular pattern act as mini cathode and mediates the reduction of H2O2 and consequently hole injection leads to oxidation-dissolution of GaN.
The I-V characteristics data is used to further calculate the Shottky barrier height value for both as grown and porous sample before and after exposing to hydrogen gas.
The forward current of all the diodes increase with exposure to hydrogen gas, which attribute to the Schottky barrier reduction caused by atomic hydrogen absorption on the metal-oxide interface.
In a constant current operation mode, this reduction of the Schottky barrier height results in a shift to lower bias voltage, which constitutes the sensor signal.
In a constant current operation mode, this reduction of the Schottky barrier height results in a shift to lower bias voltage, which constitutes the sensor signal.

Thus, corrosion may affect a RC structure inducing many effects such as loss of steel cross-sectional area, loss of steel bars ductility, concrete strength reduction, degradation of concrete cover, cover spalling and loss of steel-concrete bond strength [3-5,8,9].
The term Rc is given by the Eq. (6): (6) Methodology for lifetime prediction The proposed prediction model, defined by using Matlab software, is based on modelling both axial-flexural mechanical behaviour and corrosion-induced aging process of a RC structural element through a fibre model of the cross section subjected to maximum actions in terms of axial force-bending moment by considering the non-linear behaviour of the materials, by taking into account the following degradation-induced effects: · loss of steel cross-sectional area, evaluated through the model of Val et al. [11], which takes into account the pitting corrosion; · loss of steel bars ductility, taken into account through the reduction of the steel ultimate deformation εsu [3,12] by using the experimental model proposed by [3]; · concrete strength reduction, degradation of concrete cover and cover spalling, evaluated according to the model proposed by Coronelli and Gambarova [12].
The first step consists of providing the input data, i.e. the geometrical dimensions of the cross-section, the design life, as well as defining the PDFs of each random variable related both to model the corrosion-induced degradation process (i.e., temperature, surface chloride concentration, etc...) and mechanical properties of the materials (i.e., concrete and steel strength).
It is clear the reduction of the resistance domain over time as well as the marked discontinuity between t7=70 years and t8=80 years due to the detachment of the concrete cover.
Note that there is a reduction of the reliability index over time, increase of the failure probability, particularly for average eccentricity values, for which the moments reach higher values.

Moreover, adding the plasticizers into WG caused reduction of high intermolecular forces between protein chains, leading to increasing free volume in the protein and hence enhancing penetrating between WG and the rubber chains during blending.
On the other hand, in the case of the plasticized WG with PEG and DOP, the slight reduction of Tg was observed when PEG or DOP was added to the protein at > 10%.
In the case of the blends plasticized with glycerol and PEG, an increase in plasticizer contents from 10 to 30 wt% led to deformation of WG phase during mixing with ENR, resulting in drastical reduction of dispersed WG size.
The percent water absorption (%WA) data of WG, WG/ENR and WG/ENR/Gly, WG/ENR/PEG and WG/ENR/DOP blends containing 20 wt% of plasticizers is showed in Fig. 4.
From the DSC study, a high degree of reduction in Tg of WG was observed in glycerol plasticizer.

Two-dimensional images were obtained by raster scan of samples in x-y directions and computer recording of a complete set of spectral data for each point in the probed area.
A down-shift of the phonon Raman bands and a reduction in the LO-TO phonon frequency splitting was also observed in as-implanted samples.
The down-shift of the LO mode causes a reduction of the LO-TO splitting.
Defect-induced weakening of the long-range polarization field [7] may contribute to the reduction in the LO-TO splitting.
This mode can be used as a monitor of the carrier density in the regions where the reduction in carrier density occurs.

As a consequence, after final annealing, an increase of fraction of the grains with eta fiber orientation and a reduction of grains with gamma fiber orientation are observed, improving the magnetic induction.
The increase of hot band grain size generated a reduction on the volume fraction of undesirable component (111)[112] in the final product, due to the reduction of grain boundaries, where this orientation is nucleated [8,9] (see Fig.6).
Due to the final grain size reduction core loss presents an undesirable increase in spite of texture improvement, showing how important is the effect of the final grain size on this property. 120 140 160 180 200 220 240920 960 1000 1040 1080 1120 TEMPERATURE (°C) HBGS ( m) 0.090 0.095 0.100 0.105 0.110 0.115 0.120 0.125 0.130 (111)[[112] FRACTION HBGS (111)<112> F Fig.6: Effect of hot rolling finishing temperature on hot band grain size (HBGS) and on (111)[112] fraction in the final product.
Fig.8 confirms these affirmations for the data obtained in this research.
Despite the beneficial effect of texture improvement to reduce core loss, this property had its values deteriorated when the hot band grain size achieved values larger than 200µm, due to the reduction on final grain size caused by the increase in the amount of shear bands, as suggested by the sudden increase in Goss component fraction in the final product.