Skip to main content
. Author manuscript; available in PMC: 2022 May 25.
Published in final edited form as: Phys Biol. 2021 Jun 17;18(4):10.1088/1478-3975/abffbe. doi: 10.1088/1478-3975/abffbe

Figure 1.

Figure 1.

Comparison between conventional machine learning and deep learning. (a) In conventional machine learning, we need to extract handcrafted features from raw data. These features are used to train the classifier. (b) In deep learning, feature learning and classifier training are performed end-to-end. After the training, the trained feature extractor can produce meaningful features, which can be reused for different tasks, including unsupervised phenotyping.