ActivationCategory

Enums · AiDotNet.Enums

Categories of activation functions based on their architectural role and behavior.

Fields

Field Summary
ActivationCategory Gate Gate activations that control information flow (Sigmoid for LSTM gates, Tanh for cell state).
ActivationCategory General General-purpose activations suitable for hidden layers (ReLU, GELU, Swish).
ActivationCategory Normalization Normalization activations that produce probability distributions (Softmax, Sparsemax, LogSoftmax).
ActivationCategory Output Output activations that produce final predictions (Softmax for classification, Sigmoid for binary).
ActivationCategory Parametric Parametric activations with learnable parameters (PReLU, Maxout).
ActivationCategory Stochastic Stochastic activations with random components (GumbelSoftmax, BinarySpiking, RReLU).
int value__