Hands-On GPU Computing with Python

One-shot learning

One-shot learning can be focused on the challenging problem of the availability of a very low amount of quality drug-related datasets. Most machine learning-based techniques require training on hundreds or thousands of datasets, whereas one-shot learning aims to learn information about object categories from one or just a few training datasets (ligands).

When Graph Convolutional Neural Networks (GCNNs) are combined with one-shot learning, this significantly improves learning of meaningful distance metrics over small molecules (ligands), thus addressing the challenge faced in a traditional machine learning approach that requires a huge amount of datasets for training.

In DeepChem, molecules can be perceived mathematically ...