layer_global_attention_pool: GlobalAttentionPool

In rdinnager/rspektral: What the Package Does (One Line, Title Case)

Description

\loadmathjax

A gated attention global pooling layer as presented by Li et al. (2017).

This layer computes: \mjdeqn\boldsymbolX' = \sum\limits_i=1^N (\sigma(\boldsymbolX \boldsymbolW _ 1 + \boldsymbolb _ 1) \odot (\boldsymbolX \boldsymbolW _ 2 + \boldsymbolb _ 2))_i where \mjeqn\sigma is the sigmoid activation function.

Mode: single, disjoint, mixed, batch.

Input

• Node features of shape ([batch], N, F);

• Graph IDs of shape (N, ) (only in disjoint mode);

Output

• Pooled node features of shape (batch, channels) (if single mode, shape will be (1, channels)).

Usage

layer_global_attention_pool(
  object,
  channels,
  kernel_initializer = "glorot_uniform",
  bias_initializer = "zeros",
  kernel_regularizer = NULL,
  bias_regularizer = NULL,
  kernel_constraint = NULL,
  bias_constraint = NULL,
  ...
)

Arguments

channels	integer, number of output channels
kernel_initializer	NA
bias_initializer	initializer for the bias vectors
kernel_regularizer	regularization applied to the kernel matrices
bias_regularizer	regularization applied to the bias vectors
kernel_constraint	constraint applied to the kernel matrices
bias_constraint	constraint applied to the bias vectors.

rdinnager/rspektral documentation built on June 12, 2021, 1:26 a.m.