Description Usage Arguments See Also Examples
Construct a convolution layer.
1 2 | nnlayer_conv(layer, kernelshape, inputshape, name, inputname, stride, lowerpad,
sharing, mapcount = 1, activation = c("rlinear"), ...)
|
layer |
A layer object, e.g. using |
kernelshape |
Numeric vector describing the number of inputs in each dimension, e.g. |
inputshape |
Numeric vector describing the number of inputs in each dimension, e.g. |
name |
Name of the layer |
inputname |
Name of the preceding layer. If |
stride |
Numeric vector describing the number of inputs in each dimension, e.g. |
lowerpad |
Numeric vector describing the number of inputs in each dimension, e.g. |
sharing |
Logical vector. |
mapcount |
Number of maps to create. |
activation |
Activation function, e.g. |
... |
Other arguments passed from |
Other layer.definition.functions: nnlayer_full
,
nnlayer_input
, nnlayer_norm
,
nnlayer_output
, nnlayer_pool
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 | # Use the layer functions to generate individual layer specifications
nnlayer_input(c(13, 13))
nnlayer_input(c(3, 7, 7), name = "pixels")
# Convolution layers automatically compute the output size and padding
nnlayer_conv(NULL, c(2, 2),
inputshape = c(13, 13),
name = "conv1",
inputname = "pixels"
)
nnlayer_conv(NULL,
c(2, 2),
inputshape = c(13, 13),
name = "conv1",
inputname = "pixels",
stride = c(2, 2)
)
nnlayer_conv(NULL,
c(1, 2, 2),
inputshape = c(3, 13, 13),
name = "conv1",
inputname = "pixels",
stride = c(1, 2, 2)
)
nnlayer_pool(NULL,
c(1, 2, 2),
inputshape = c(3, 13, 13),
name = "conv1",
inputname = "pixels",
stride = c(1, 2, 2)
)
# Specify the number of nodes in a fully connected layer
nnlayer_full(NULL, nodes = 100, name = "h3", inputname = "conv")
# Output layer
nnlayer_output(NULL, 6, name = "class", inputname = "h3")
# using magrittr pipes to connect layers ----------------------------------
require(magrittr)
nnlayer_input(c(3, 50, 50), name = "pixels") %>%
nnlayer_conv(
kernelshape = c(1, 5, 5),
name = "conv1",
stride = c(1, 2, 3)
)
nnlayer_input(c(3, 50, 50), name = "pixels") %>%
nnlayer_conv(
kernelshape = c(1, 5, 5),
name = "conv1",
stride = c(1, 2, 3)
) %>%
nnlayer_pool(
kernelshape = c(1, 5, 5),
name = "conv1",
stride = c(1, 2, 3)
)
nnlayer_norm(NULL, inputshape = c(3, 11, 5), kernelshape = c(1,5,5), name = "rnorm1", inputname = "conv")
nnlayer_input(c(3, 50, 50), name = "pixels") %>%
nnlayer_conv(
kernelshape = c(1, 5, 5),
name = "conv1",
stride = c(1, 2, 3)
) %>%
nnlayer_norm(
kernelshape = c(1, 5, 5),
name = "norm1",
stride = c(1, 2, 3),
alpha = 0.0001,
beta = 0.75
)
nnlayer_input(c(3, 50, 50), name = "pixels") %>%
nnlayer_conv(
kernelshape = c(3, 5, 5),
name = "conv1",
stride = c(1, 2, 2),
mapcount = 48
) %>%
nnlayer_conv(
kernelshape = c(1, 4, 4),
stride = c(1, 2, 2),
name = "conv2"
) %>%
nnlayer_full(nodes = 100, name = "hid1") %>%
nnlayer_full(nodes = 30, name = "hid2") %>%
nnlayer_output(nodes = 6, name = "class")
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.