modules.conv.DecoderNet1d¶
net = mdnc.modules.conv.DecoderNet1d(
channel, layers, out_size,
kernel_size=3, in_length=2, out_planes=1
)
This moule is a built-in model for 1D convolutional decoder network. This network could be used as a part of the auto-encoder, or just a network for up-sampling (or generating) data.
The network would up-sample the input data according to the network depth. The depth is given by the length of the argument layers. The network structure is shown in the following chart:
flowchart TB
u1["Block 1<br>Stack of layers[0] layers"]
u2["Block 2<br>Stack of layers[1] layers"]
ui["Block ...<br>Stack of ... layers"]
un["Block n<br>Stack of layers[n-1] layers"]
optional:::blockoptional
subgraph optional [Optional]
cin["Conv1d<br>with unsqueeze"]
end
u1 -->|up<br>sampling| u2 -->|up<br>sampling| ui -->|up<br>sampling| un
cin -.-> u1
linkStyle 0,1,2 stroke-width:4px, stroke:#080 ;
classDef blockoptional fill:none, stroke-dasharray:10,10, stroke:#9370DB, width:100;The argument layers is a sequence of int. For each block \(i\), it contains layers[i-1] repeated modern convolutional layers (see mdnc.modules.conv.ConvModern1d). Each up-sampling is configured by stride=2. The channel number would be doubled in the up-sampling route. An optional unsqueezer and convolutional layer could be prepended to the first layer when the argument in_length != None. This optional layer is used for converting the vector features in initial feature maps.
Arguments¶
Requries
| Argument | Type | Description |
|---|---|---|
channel | int | The channel number of the first hidden block (layer). After each down-sampling, the channel number would be doubled. |
layers | (int,) | A sequence of layer numbers for each block. Each number represents the number of convolutional layers of a stage (block). The stage numer, i.e. the depth of the network is the length of this list. |
out_size | int | The length of the output data. This argument needs to be specified by users, because the network needs to configure its layers according to the output size. |
kernel_size | int | The kernel size of each convolutional layer. |
in_length | int | The length of the input vector, if not set, the input needs to be feature maps. See the property input_size to check the input data size in this case. |
out_planes | int | The channel number of the output data. |
Operators¶
__call__¶
y = net(x)
The forward operator implemented by the forward() method. The input data is a tensor with a size determined by configurations. The output is a 1D tensor. The channel number of the output is specified by the argument out_planes.
Requries
| Argument | Type | Description |
|---|---|---|
x | torch.Tensor | A tensor,
|
Returns
| Argument | Description |
|---|---|
y | A 1D tensor, the size should be (B, C, L), where B is the batch size, C is the input channel number, and L is the output data size specified by the argument out_size. |
Properties¶
nlayers¶
net.nlayers
The total number of convolutional layers along the depth of the network. This value would not take the fully-connected layer into consideration.
input_size¶
net.input_size
The size of the input data size (a tuple). This property is useful when in_length is None. In this case, the input size is determined by the network.
Warning
This size contains the channel number (as the first element), because the input channel number is also determined by network when in_length is None.
Examples¶
Example 1
1 2 3 4 5 6 | |
The number of convolutional layers along the depth is 18.
The input size is (32,).
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv1d-1 [-1, 1024, 4] 132,096
Conv1d-2 [-1, 1024, 4] 3,145,728
InstanceNorm1d-3 [-1, 1024, 4] 2,048
PReLU-4 [-1, 1024, 4] 1,024
Conv1d-5 [-1, 512, 4] 1,572,864
_ConvModernNd-6 [-1, 512, 4] 0
InstanceNorm1d-7 [-1, 512, 4] 1,024
PReLU-8 [-1, 512, 4] 512
Conv1d-9 [-1, 512, 4] 786,432
_ConvModernNd-10 [-1, 512, 4] 0
InstanceNorm1d-11 [-1, 512, 4] 1,024
PReLU-12 [-1, 512, 4] 512
Upsample-13 [-1, 512, 8] 0
Conv1d-14 [-1, 512, 8] 786,432
_ConvModernNd-15 [-1, 512, 8] 0
_BlockConvStkNd-16 [-1, 512, 8] 0
InstanceNorm1d-17 [-1, 512, 8] 1,024
PReLU-18 [-1, 512, 8] 512
Conv1d-19 [-1, 256, 8] 393,216
_ConvModernNd-20 [-1, 256, 8] 0
InstanceNorm1d-21 [-1, 256, 8] 512
PReLU-22 [-1, 256, 8] 256
Conv1d-23 [-1, 256, 8] 196,608
_ConvModernNd-24 [-1, 256, 8] 0
InstanceNorm1d-25 [-1, 256, 8] 512
PReLU-26 [-1, 256, 8] 256
Upsample-27 [-1, 256, 16] 0
Conv1d-28 [-1, 256, 16] 196,608
_ConvModernNd-29 [-1, 256, 16] 0
_BlockConvStkNd-30 [-1, 256, 16] 0
InstanceNorm1d-31 [-1, 256, 16] 512
PReLU-32 [-1, 256, 16] 256
Conv1d-33 [-1, 128, 16] 98,304
_ConvModernNd-34 [-1, 128, 16] 0
InstanceNorm1d-35 [-1, 128, 16] 256
PReLU-36 [-1, 128, 16] 128
Conv1d-37 [-1, 128, 16] 49,152
_ConvModernNd-38 [-1, 128, 16] 0
InstanceNorm1d-39 [-1, 128, 16] 256
PReLU-40 [-1, 128, 16] 128
Upsample-41 [-1, 128, 32] 0
Conv1d-42 [-1, 128, 32] 49,152
_ConvModernNd-43 [-1, 128, 32] 0
_BlockConvStkNd-44 [-1, 128, 32] 0
InstanceNorm1d-45 [-1, 128, 32] 256
PReLU-46 [-1, 128, 32] 128
Conv1d-47 [-1, 64, 32] 24,576
_ConvModernNd-48 [-1, 64, 32] 0
InstanceNorm1d-49 [-1, 64, 32] 128
PReLU-50 [-1, 64, 32] 64
Conv1d-51 [-1, 64, 32] 12,288
_ConvModernNd-52 [-1, 64, 32] 0
InstanceNorm1d-53 [-1, 64, 32] 128
PReLU-54 [-1, 64, 32] 64
Upsample-55 [-1, 64, 64] 0
Conv1d-56 [-1, 64, 64] 12,288
_ConvModernNd-57 [-1, 64, 64] 0
_BlockConvStkNd-58 [-1, 64, 64] 0
InstanceNorm1d-59 [-1, 64, 64] 128
PReLU-60 [-1, 64, 64] 64
Conv1d-61 [-1, 64, 64] 12,288
_ConvModernNd-62 [-1, 64, 64] 0
InstanceNorm1d-63 [-1, 64, 64] 128
PReLU-64 [-1, 64, 64] 64
Conv1d-65 [-1, 64, 64] 12,288
_ConvModernNd-66 [-1, 64, 64] 0
InstanceNorm1d-67 [-1, 64, 64] 128
PReLU-68 [-1, 64, 64] 64
Upsample-69 [-1, 64, 128] 0
Conv1d-70 [-1, 64, 128] 12,288
_ConvModernNd-71 [-1, 64, 128] 0
_BlockConvStkNd-72 [-1, 64, 128] 0
Conv1d-73 [-1, 3, 128] 963
DecoderNet1d-74 [-1, 3, 128] 0
================================================================
Total params: 7,505,667
Trainable params: 7,505,667
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.00
Forward/backward pass size (MB): 1.88
Params size (MB): 28.63
Estimated Total Size (MB): 30.51
----------------------------------------------------------------
Example 2
1 2 3 4 5 6 | |
The number of convolutional layers along the depth is 17.
The input size is (1024, 4).
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
Conv1d-1 [-1, 1024, 4] 3,145,728
InstanceNorm1d-2 [-1, 1024, 4] 2,048
PReLU-3 [-1, 1024, 4] 1,024
Conv1d-4 [-1, 512, 4] 1,572,864
_ConvModernNd-5 [-1, 512, 4] 0
InstanceNorm1d-6 [-1, 512, 4] 1,024
PReLU-7 [-1, 512, 4] 512
Conv1d-8 [-1, 512, 4] 786,432
_ConvModernNd-9 [-1, 512, 4] 0
InstanceNorm1d-10 [-1, 512, 4] 1,024
PReLU-11 [-1, 512, 4] 512
Upsample-12 [-1, 512, 8] 0
Conv1d-13 [-1, 512, 8] 786,432
_ConvModernNd-14 [-1, 512, 8] 0
_BlockConvStkNd-15 [-1, 512, 8] 0
InstanceNorm1d-16 [-1, 512, 8] 1,024
PReLU-17 [-1, 512, 8] 512
Conv1d-18 [-1, 256, 8] 393,216
_ConvModernNd-19 [-1, 256, 8] 0
InstanceNorm1d-20 [-1, 256, 8] 512
PReLU-21 [-1, 256, 8] 256
Conv1d-22 [-1, 256, 8] 196,608
_ConvModernNd-23 [-1, 256, 8] 0
InstanceNorm1d-24 [-1, 256, 8] 512
PReLU-25 [-1, 256, 8] 256
Upsample-26 [-1, 256, 16] 0
Conv1d-27 [-1, 256, 16] 196,608
_ConvModernNd-28 [-1, 256, 16] 0
_BlockConvStkNd-29 [-1, 256, 16] 0
InstanceNorm1d-30 [-1, 256, 16] 512
PReLU-31 [-1, 256, 16] 256
Conv1d-32 [-1, 128, 16] 98,304
_ConvModernNd-33 [-1, 128, 16] 0
InstanceNorm1d-34 [-1, 128, 16] 256
PReLU-35 [-1, 128, 16] 128
Conv1d-36 [-1, 128, 16] 49,152
_ConvModernNd-37 [-1, 128, 16] 0
InstanceNorm1d-38 [-1, 128, 16] 256
PReLU-39 [-1, 128, 16] 128
Upsample-40 [-1, 128, 32] 0
Conv1d-41 [-1, 128, 32] 49,152
_ConvModernNd-42 [-1, 128, 32] 0
_BlockConvStkNd-43 [-1, 128, 32] 0
InstanceNorm1d-44 [-1, 128, 32] 256
PReLU-45 [-1, 128, 32] 128
Conv1d-46 [-1, 64, 32] 24,576
_ConvModernNd-47 [-1, 64, 32] 0
InstanceNorm1d-48 [-1, 64, 32] 128
PReLU-49 [-1, 64, 32] 64
Conv1d-50 [-1, 64, 32] 12,288
_ConvModernNd-51 [-1, 64, 32] 0
InstanceNorm1d-52 [-1, 64, 32] 128
PReLU-53 [-1, 64, 32] 64
Upsample-54 [-1, 64, 64] 0
Conv1d-55 [-1, 64, 64] 12,288
_ConvModernNd-56 [-1, 64, 64] 0
_BlockConvStkNd-57 [-1, 64, 64] 0
InstanceNorm1d-58 [-1, 64, 64] 128
PReLU-59 [-1, 64, 64] 64
Conv1d-60 [-1, 64, 64] 12,288
_ConvModernNd-61 [-1, 64, 64] 0
InstanceNorm1d-62 [-1, 64, 64] 128
PReLU-63 [-1, 64, 64] 64
Conv1d-64 [-1, 64, 64] 12,288
_ConvModernNd-65 [-1, 64, 64] 0
InstanceNorm1d-66 [-1, 64, 64] 128
PReLU-67 [-1, 64, 64] 64
Upsample-68 [-1, 64, 128] 0
Conv1d-69 [-1, 64, 128] 12,288
_ConvModernNd-70 [-1, 64, 128] 0
_BlockConvStkNd-71 [-1, 64, 128] 0
Conv1d-72 [-1, 3, 128] 963
DecoderNet1d-73 [-1, 3, 128] 0
================================================================
Total params: 7,373,571
Trainable params: 7,373,571
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.02
Forward/backward pass size (MB): 1.85
Params size (MB): 28.13
Estimated Total Size (MB): 29.99
----------------------------------------------------------------
Last update: March 14, 2021