Compression of an Infrared Spectral Library by Karhunen-Loeve Transformation

Abstract

A data compression technique using factor analysis and the Karhunen-Loeve transformation has been applied to a library of vapor phase infrared spectra. Application of the compression algorithm results in a fivefold reduction in storage requirements for the spectral library and a corresponding reduction in search time. A search system based on the compression data has been evaluated and found to compare favorably with a search using the entire spectrum.

A neural network approach to infrared spectrum interpretation

“…Many artificial intelligence methods have been tried, including simple search and matching procedures [2,3]; the use of correlation tables [4][5][6][7][8] and correlation coefficients [93; expert system approaches, both rule-driven [10][11][12][13][14][15][16][17][18][19][20][21][22] and table-driven [23]; the use of Bayesian statistics [24,25] and other statistical approaches [26,27]; set theory [28], including fuzzy sets [2,29,30]; and binary linear separation [31][32][33][34][35][36], KNN separation [25,36,37], principal component analysis [381, eigenvector projection [39], hierarchical cluster analysis [40][41][42], factor analysis [39], and other pattern recognition techniques [43][44][45][46][47]. Many artificial intelligence methods have been tried, including simple search and matching procedures [2,…”

Application of robust eigenvectors to the compression of infrared spectral libraries

“…The compression of spectral libraries by eigenvector projection has been described previously in the literature (1)(2)(3)5). An excellent introduction and review of the application of eigenvector methods applied to chemistry is found in ref 6.…”

“…Fluorescence line narrowing (FLN) spectrometry has recently been shown to possess the attributes required for the analysis of cellular macromolecular (e.g., DNA, globin) damage resulting from covalent binding between the macromolecule and carcinogenic metabolites of polycyclic aromatic hydrocarbons (PAHs) (1). The formation of a DNA adduct is generally believed to be a crucial step in the pathway that leads to carcinogenesis and tumorigenesis (2)(3)(4).…”

“…Compression of spectra by eigenvector projection removes redundant information, which makes it particularly well suited for Fourier transform infrared (FT-IR) spectra (1)(2)(3). This technique retains discriminatory information regarding band location, intensity, and shape.…”

Using analytical data to build expert systems

“…Numerical pattern recognition techniques have been widely used for this task. [1][2][3][4] These numerical techniques frequently seek to cluster the data by transforming the inherent data attributes to new and 0095-2338/88 /1628-0159S01.50/0 possibly composite attributes. This has the conceptual disadvantage that the new attributes may be very difficult to relate to the original data from which they were derived.…”