Representation of the Objects

the objects of investigation, chemical compounds or chemical reactions, have to be represented. Chemical compoimds wUl mostly be represented by their molecular structure in various forms of sophistication. This task is addressed in Chapter 2. The representation of chemical reactions is dealt with in Chapter 3. The vast number of compounds known can only be managed by storing them 

The two ways of learning - deductive and inductive - have already been mentioned. Quite a few properties of chemical compounds can be calculated explicitly. Foremost of these are quantum mechanical methods. However, molecular mechanics methods and even simple empirical methods can often achieve quite high accuracy in the calculation of properties. These deductive methods are discussed in Chapter 7. 

Inductive methods for establishing a correlation between chemical compounds and their properties are the theme of Chapter 9. In many cases, the structure of chemical compounds has to be pre-processed in order to make it amenable to inductive learning methods. This is usually achieved by means of structure descriptors, methods for the calculation of which are outlined in Chapter 8. 

After approaches to the solution of the major tasks in chemoinformatics have thus been outlined, these methods are put to work in specific applications. Some of these apphcations, such as structure elucidation on the basis of spectral information, reaction prediction, computer-assisted synthesis design or drug design, are presented in Chapter 10. 

---

The intention of this chapter has been to provide an overview of analytical methods for predicting and reducing human error in CPI tasks. The data collection methods and ergonomics checklists are useful in generating operational data about the characteristics of the task, the skills and experience required, and the interaction between the worker and the task. Task analysis methods organize these data into a coherent description or representation of the objectives and work methods required to carry out the task. This task description is subsequently utilized in human error analysis methods to examine the possible errors that can occur during a task. 

Fig. 44.8. (a) The structure of the neural network, for solving the XOR classification problem of Fig. 44.7. (b) The two boundary lines as defined by the hidden units in the input space (jtl, 2). (c) Representation of the objects in the space defined by the output values of the two hidden units ( hul, hu2) and the boundary line defined in this space by the output unit. The two objects of class A are at the same location. 

This technique functions by taking observed measures of similarity or dissimilarity between every pair of M objects, then finding a representation of the objects as points in Euclidean space so that the interpoint distances in some sense match the observed similarities or dissimilarities by means of weighting constants. 

When the significant eigenvectors are more than 2 or 3, the information cannot be easily visualized by few eigenvector plots. In these cases the use of nonlinear mapping (NLM) can give a planar representation of the objects with greater fidelity to the structure of the information in the hyperspace of the variables... 

Confronted with a problem in which two data sets were available, Breedlove et al. (1977) chose a solution that minimizes a sum of terms not unlike expression (56). Available were two images one a blurred representation of the object, the other a superposition of sharp renderings. In this sum, the right-hand term accommodates the blurred image as in expression (56). The other term incorporates the multiple exposure via the Lagrange multiplier technique. Solutions obtained by this method illustrated the desirability of using all the available data. 

If one wishes to represent the objects in the space of the factors, one has to calculate the matrix of factor scores F. The procedure is a multiple regression between the original values and the factors and is also called estimation according to BARTLETT [JAHN and YAHLE, 1970]. The graphical representation of the objects may be used to detect groups of related objects or to identify objects which are strongly related to one or more of the factors. 

Figure 5.1 Schematic presentation of town Konigsberg with its seven bridges and connections. The plan of the town (a) is converted into graphs (b and c), which are completely equivalent representation of the object in (5.1a) and between themselves in Graph Theory.

The amount of information is gradually withdrawn from the pattern as the number of coefficients for the back-transformation is reduced. The zero order coefficients, co of both transforms, Fourier and Hadamard, carry the sum (integral) of all elements of the original representation of the object and do not contribute any other information (ref. 5). Figure 5.2 shows the order of importance of other coefficients for both transformations. 

A common representation of the objective function in Eq. (3) minimizes the mean-square error of approximation as... 

Assignment of features to facial parts leads to representation of the objects as faces. Well known are the faces introduced by Chernoff. The features are characterized by facial parameters, such as the size or curvature of the eyes, the mouth, the eye brows, the nose, or the upper and lower half of the case. As an example, the hair data of the three subject types are plotted in Figure 5.22C, as Chernoff faces. 

Fig. 4.6 A flowchart representation of the objective tree as presented by Team 1...

A software system processes raw point clouds or volumetric data and transfers them into a virtual representation of the object such as surfaces and features. One of the critical tasks of vision-based manufacturing apphcations is to generate a virtual representation and its success rehes on reliable algorithms and tools. Processing of raw scanned data or data cleaning is very important since curves and reconstmcted surfaces are based on the mesh model. Data processing and surface reconstmction is the centre piece of a RE process. The interpretation of raw data to a required computer model is a complicated process, and it involves the following typical issues [10] ... 

Aberration A perfect lens would produce an image that was a scaled representation of the object real lenses suffer from defects known as aberrations and measured by aberration coefficients. 

Interventions are description-dependent insofar as they are intended actions that are based on specific explicit representations of the object we intervene on. Necessarily, any intervention on water is an intervention on H2O, and vice versa. However, an intended intervention can, intuitively, be construed as an intervention that is relative to a certain representation of the object intervened on. Let an intended intervention be an intervention on x that is planned with respect to a representation by a property structure that presents us with x. The hyper-intensionality is here inherited from the hyper-intensionality of intentionaUty. If I intend to intervene upon the temperature of a piece of ice, I do not thereby intend to intervene upon the kinetic energy and the lattice structures of the sum of corresponding H2O molecules. Intended interventions are always description-, and, hence, property-structure-relative. 

Based on these multivariate distances or similarities, clusters of objects are generated. The distance between single objects within the clusters is minimized while the distance between clusters is maximized. The objective of the procedure is a clear representation of the objects with fewer dimensions. Cluster analysis is a multivariate explorative method which needs no a priori information and yields no statistical evidence about group memberships. 

---