Figure 2
Graphical overview for the classification submodule. The PathMNIST dataset is used to train a ResNet-18 network based on different initial conditions: randomized initial weights and biases (first experiment), pre-trained network weights and biases that are frozen during training (second experiment) and pre-trained network weights and biases that are unfrozen during training (third experiment). Following network training, the prediction accuracy of the three networks is compared.

Graphical overview for the classification submodule. The PathMNIST dataset is used to train a ResNet-18 network based on different initial conditions: randomized initial weights and biases (first experiment), pre-trained network weights and biases that are frozen during training (second experiment) and pre-trained network weights and biases that are unfrozen during training (third experiment). Following network training, the prediction accuracy of the three networks is compared.

Close
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close

This PDF is available to Subscribers Only

View Article Abstract & Purchase Options

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Close