Fig. 1
From: DeepSAT: Learning Molecular Structures from Nuclear Magnetic Resonance Data
Overview of DeepSAT. a Web-based platform of DeepSAT analysis (https://deepsat.ucsd.edu). b The multi-task learning architecture of DeepSAT. In the feature extraction step, the convolutional neural network extracts the features from HSQC spectra. Based on the extracted features, fully connected layers predict Morgan fingerprints, molecular weights, and chemical classes. By using the predicted properties, structure annotation is accelerated