Hadamard transformed carbon-13 nuclear magnetic resonance spectra. Pattern recognition analysis

Abstract

Carbon-13 nuclear magnetic resonance spectra represented either in binary coded (presence or absence of peaks) or relative intensity form have been subjected to Hadamard Transform processing. Results of linear discriminant function analysis of both Hadamard sequency ordered carbon-13 spectra and Hadamard energy spectra are described. These results are compared to those obtained in a previous study in which Hadamard preprocessing was not used. It is shown that learning machine analysis of the Hadamard domain data is, in general, less efficacious.Recently, we reported (1) the first application of the heuristic pattern recognition method called the "learning machine" (2) approach to the analysis of experimental proton noise-decoupled 13C high resolution nuclear magnetic resonance spectra. We found that the general applicability of 1 Author to whom requests for reprints should be sent.

Evaluation and comparison of pattern classifiers for chemical applications

“…The four quantities TV, TVpred, Arcorr, and TV,otai are sufficient to compute all the relevant probabilities defined previously. p(l) = TV/TVtota, (16) p(2) = 1 -P(l) = (TVtotai -TV)/yvtotal (17) p(j) = TVpred/7Vtotal (18) P(n) = 1 -p(j) = (TVtotai ~TVpred)/TVtotai (19) p(j|l) = 7Vcorr/7V (20) p(Lj) =p(j|l)p(l) = TVcorr/TVtotal (21) p(l|j) =p(lj)/p(j) = TVcorr/TVpred (22) P(Ln) = p(l) -p(l,j) = (TV -TV"")/TVto,al (23) P(2J) = P(j) -P( 1 J) = (Apred -7Vcorr)/TVtota| (24) p(2,n) = p(2) -p(2,j) P(2,n) . (Atotal -A -TVpred + TVcorr) (27) p ( 2) (Atotal -A)…”

Classification of binary carbon-13 nuclear magnetic resonance spectra

“…In previous papers, investigations of a number of heuristic pattern recognition strategies for the computer-assisted interpretation of nuclear magnetic resonance spectra were reported (1)(2)(3)(4)(5). The impetus for this work was our belief that the sensitivity of I3C chemical shifts to structure variation and the widespread availability of instrumentation for routine natural abundance 13C spectral measurements made this structure elucidation technique an imperative target of pattern recognition research.…”

“…49, NO. 14, DECEMBER 1977 Earlier work had relied, for training of categorizers, exclusively on a set of 500 spectra drawn from the Johnson and Jankowski collection (6) and was primarily devoted to studies of the merits of various data preprocessing techniques (1)(2)(3). In this paper, results made possible by the use of a new and larger carbon-13 NMR data base containing close to four thousand "C NMR spectra are reported.…”

“…As a result, a number of such techniques have been the subject of recent investigation by both ourselves and others. These methods fall into the three broad categories of: parametric methods (which, in general, have relied on the development of linear additivity parameters) (14-19); heuristic methods (1)(2)(3)(4)20); and the use of similarity measures (of which, we consider spectral search systems to be one example) (12,14,21). The present results bear on the relative efficacy of learning machine (heuristic) interpretation of spectra as opposed to use of empirical peak occurrence probabilities for structural characterization via either Bayes or maximum likelihood classifiers.…”

Computer-assisted interpretation of carbon-13 nuclear magnetic resonance spectra applied to structure elucidation of natural products

“…The enormous sensitivity of 13C chemical shifts to structural changes is making 13C-NMR spectroscopy a valuable tool in structure elucidation work (10,11). Wilkins and co-workers (12)(13)(14)(15) have reported encouraging results using linear learning machines and a committee threshold logic unit on 13C-NMR spectra. The present investigation has a twofold purpose.…”