Using the simulated spectra we developed, we trained a new neural network search engine for the xyloglucan database. The purpose of this project was to see if a neural network that has only seen artificially simulated spectra of a certain group of compounds can identify the compounds when presented with the real spectra of these compounds. We were able to train a neural network which could recognize the real spectra. This capability enables us to bypass the bottleneck of data collection for neural network training. We have installed the new search engine on the CCRC-Net system on the WWW.
Showing the most recent 10 out of 245 publications