Spatial sample representation

In order to better understand supervised learning methods, it is useful to recall the concept of spatial sample representation which was mentioned earlier in Section 8.2.7. The development of any supervised learning method involves three steps ... 

When a powerful la.ser pulse is focused on a 5- to 30-pm spot on a sample surface, a small amount of the solid is vaporized regardless of whether it is a conductor. The resulting plume is made up of atoms, ions, and molecules. In the microprobe, the contents of the plume arc excited by a spark between a pair of small electrodes located immediately above the surface of the sample. The re.sulting radiation is then focused on a suitable monochromator-detection system. With this type of source, it has been possible to determine the trace element composition of single blood cells and tiny inclusion areas in alloys. The laser can be scanned across a surface to obtain a spatially resolved representation of surface composition. 

Fig. 19a, b. Curves of isometric heating a reference sample (curve 1) and sample obtained if molecular orientation exists and containing a spatial framework (curve 2) (schematic representation) b data for high density polyethylene (—A----------ordinary fiber, —O-------fiber obtained by... 

Fig. 21 a-c. Schematic representation supramolecular structure of a crystalline rigid-chain polymer (a), an idealized ECC of a flexible-chain polymer (b) and an orientationally crystallized sample with a spatial ECC framework (c)... 

Figure 5.18. Schematic representation of the density of states N(E) in the conduction band and of the definitions of work function d>, chemical potential of electrons p, electrochemical potential of electrons or Fermi level p, surface potential x> Galvani (or inner) potential

Method Amount of tissue required Spatial resolution Requirements for sample preparation Lower detection limit (approximate) Pictorial representation Quantification possible Precision... 

Figure 12.15 Spatial representation of a set of calibration samples (denoted by small circles) and two unknown samples (denoted by large circles). The mean of the calibration samples is at the origin.

It is extremely useful to move beyond a subjective and qualitative analysis of the spatial distribution of sample components, and to begin to explore the quantitative information contained within chemical imaging data sets. One of the most powerful statistical representations of an image does not even maintain spatial information. A chemical image can be represented as a histogram, with intensity along the x-axis and the number of pixels with that intensity along the y-axis. This is a statistical and quantitative... 

Such spatial representation of single-data profiles provides a very powerful view of the data, where all samples in a data set can be readily compared to one another. Figure 8.9 shows a more conventional profile display of analytical data obtained from five different... 

Cartesian coordinates are a convenient alternative representation for a spatial distribution function. Being uniform over the local space, the data structure obtained is easy to represent (access), to normalize, and to visualize. Use of a Cartesian representation becomes a necessity for complex or very flexible molecules. The principal drawbacks of this coordinate system are the size of the data structure it generates (typically about 1,000,000 elements), its inherent inefficiency (since the grid size is determined by the shortest dimension of the smallest feature one hopes to capture), and the fact that its sampling pattern is usually not commensurate with the structures one wants to represent (which can cause artificial surface features or textures when visualized). Obtaining sufficiently well-averaged results in more distant volume elements can be a problem if the examination of more subtle secondary features is desired. See Figures 7, 8 and 9 for examples of SDFs that have utilized Cartesian coordinates. 

All three estimates of A based on measurements of free tropospheric or canopy air were close to each other (Table 3). Since the spatial representation of the flask-sampling networks is still limited (Tans et al, 1996), the collection of canopy air to deduce A, can be recommended. 

---