Beginning Anomaly Detection Using

rate: A float between 0 and 1 that determines the proportion of input 

units to drop.

• 

data_format: ‘channels_first’ or ‘channels_last’. This tells the 

flattening layer how to format the flattened output to preserve the 

formatting of channels first or channels last.

 Conv1D

keras.layers.Conv1D()

Check out Chapter 

7

 for a detailed explanation on how one-dimensional 

convolutions work.

This layer is a one-dimensional (or temporal) convolutional layer. It basically passes 

a filter over the one-dimensional input and multiplies the values element-wise to create 

the output feature map.

These are the parameters that the function takes:

• 

filters: An integer value that determines the dimensionality of the 

output space. In other words, this is also the number of filters in the 

convolution.

• 

kernel_size: An integer (or tuple/list of a single integer) that specifies 

the length of the filter/kernel that is used in the 1D convolution.

• 

strides: An integer (or tuple/list of a single integer) that tells the 

layer how many data entries to shift by after one element-wise 

multiplication of the filter and the input data. Note: A stride value != 1 

isn’t compatible if the dilation_rate != 1.

Appendix A   intro to KerAs

---

335

• 

padding: ‘valid’, ‘causal’, or ‘same’. ‘valid’ doesn’t zero pad the output.  

‘same’ zero pads the output so that it’s the same length as the input. 

‘causal’ padding generates causal, dilated convolutions. For an 

explanation on what ‘causal’ padding is, refer to Chapter 

7

.

• 

data_format: ‘channels_first’ or ‘channels_last’. This tells the 

flattening layer how to format the flattened output to preserve the 

formatting of channels first or channels last. ‘channels_first’ has 

the format (batch, features, steps), and ‘channels_last’ has the 

format (batch, steps, features).

• 

dilation_rate: An integer (or tuple/list of a single integer) serves 

as the dilation rate for this dilated convolutional layer. For an 

explanation of how this works, refer to Chapter 

7

.

•