Skip to main content
. Author manuscript; available in PMC: 2019 Jul 15.
Published in final edited form as: Anal Chem. 2019 Apr 1;91(8):5191–5199. doi: 10.1021/acs.analchem.8b05821

Figure 1.

Figure 1.

Comparison between DNN and classical machine learning for CCS prediction. Blue sections are purely computational, making DNN an almost completely computational approach. Classical machine learning requires an input of well-defined and comprehensive molecular descriptors, which can be adversely influenced by domain knowledge (e.g., CCS is correlated with the m/z value), reducing its accuracy.