Understanding the patterns that neural networks learn from chemical spectra
Laura Rieger, Max L. Wilson, Tejs Vegge, Eibar Flores
Abstract
We train a convolutional neural network to classify functional groups from infrared spectra. With explainability methods, we show the model uses the presence and absence of peaks, at fundamental and anharmonic frequencies for accurate classification.
Topics & Concepts
Artificial neural networkComputer scienceArtificial intelligenceMetabolomics and Mass Spectrometry StudiesComputational Drug Discovery MethodsFault Detection and Control Systems