Preprocessing Filters

Finally, we employed a list of 56 substructures that are either required or forbidden to be in compounds we select. These represent decades of medicinal chemistry 

---

In the Dataset menu the columns (called panels by the software) are chosen for preprocessing. Because the software has automatically excluded the identifier columns, all columns are chosen, and the icon Add is clicked. The default thresholds are accepted by clicking Add. Under row variation, a maximum/ minimum of two and a maximum minus minimum of 100 is chosen. This sets the filter at twofold expression change and 100 above background. The icon Add is clicked. The data are normalized to have mean = 0 and variance =1... 

Fig. 13.6 Preprocessing compound files. Representative substructural exclusion criteria and compound filters used to prune the CRCD database are shown. These were designed to eliminate reactive compounds, toxic materials, and so on. The complete list of exclusion criteria was developed over a number of years in close collaboration with experienced medicinal chemists.

This filtering preprocessing method can be used whenever the variables are expressed as a continuous physical property. One example is dispersive or Fourier-Transform spectral data, where the spectral variables refer to a continuous series of wavelength or wavenumber values. In these cases, derivatives can serve a dual purpose (I) they can remove baseline offset variations between samples, and (2) they can improve the resolution of overlapped spectral features. 

Descriptor matrix coordinates can be preprocessed using the recently developed filtering technique (37) known as OSC, which is also implemented in... 

Thus the data can be preprocessed by Fourier transforming V( ) 2, applying a window corresponding to the extent of the pupil function of the lens, and then inverse transforming to obtain a filtered V(m), which can then be used as the data for the Gerchberg-Saxton algorithm. 

Chemometrics finds widespread use in spectroscopy, and there are a number of reviews that describe advances in this area. In a review by Geladi [41], some of the main methods of chemometrics are illustrated with examples. A series of three reviews addresses the topic of chemometrics in spectroscopy [42-44], Part 1 has 199 references and focuses specifically on chemometric techniques applied to spectroscopic data [42], Part 2 has 68 references and focuses on data-preprocessing methods and data transformations aimed at reducing noise, removing effects of baseline offsets, and filtering to remove noise [43], Part 3 focuses on multiway methods of analysis applied to spectroscopic data [44],... 

Process data for the same polymer recipe were analyzed for 50 noncon-secutive, sequential batches. As before, the data were preprocessed to remove outliers and sorted to reflect normal operation. The data are sampled at 1-minute intervals during production of each batch. Filtering and normalization of the data is done prior to analysis. The final polymer quality... 

Figure 12.8 shows the results of a feasibility experiment carried out in transmission Raman geometry on a standard paracetamol (acetaminophen) tablet, with and without a bandpass filter. The bandpass filter was centered at 830 nm and its bandwidth was 3.2 nm. The Raman spectra are shown in a raw form, without any numerical preprocessing. A substantial enhancement (x6.5) was observed upon the insertion of the bandpass filter into the proximity of the tablet. Such a large enhancement could be very difficult, and in many cases impossible, to reproduce by other means. The use of a bandpass filter permits the collection of Raman spectra of higher quahty, and consequently a more sensitive analysis of sample can be accompHshed, or higher penetration depths attained. It is also worth noting... 

The aim of the signal analysis in the MErKoFer project was to determine significant deviations of the various signals (i.e., process and other recorded values) from their stable state. Of course, care had to be taken of noise, oscillations, and similar issues. Several methods where therefore applied and evaluated. Partially, Gauss or median filters were used for preprocessing. 

There are many educational and commercial software packages available for development and deployment of ANNs. Some of those packages such as Gensym s NeurOn-Line Studio include data preprocessing modules to filter or scale data and eliminate outliers [89]. 

---