layer_kernel_dot: Perform Kernel Dot Product

In tpq/caress: Caress, a Gentler Introduction to Keras

Description

This function calls k_dot as a "lambda" layer.

Usage

layer_kernel_dot(layer, kernel, name = NULL)

Arguments

layer	The incoming layer.
kernel	A layer that represents a kernel.
name	A string. The prefix label for all layers.

Value

The result of k_dot.

Examples

library(keras)
library(caress)
set.seed(1)
index <- sample(1:nrow(iris))
x <- as.matrix(iris[index,1:4])
y <- to_categorical(as.numeric(iris[index,5])-1)
k_clear_session()
input <- from_input(x)
weight <- input %>%
  layer_learnable_array(4) %>%
  layer_reshape(c(4,1))
target <- layer_kernel_dot(input, weight) %>%
  layer_flatten() %>%
  to_output(y)
m <- prepare(input, target)
build(m, x, y, batch_size = 4, epochs = 100)

tpq/caress documentation built on March 11, 2021, 8:03 p.m.