Beginning Anomaly Detection Using

two-dimensional convolutional layer.

This is a standard convolution, equivalent to what you looked at in Chapter 

3

.  

You can also think of a standard convolutional layer as having a dilation factor of 

one 

(refer to Figure 

7-7

).

Now, let’s look at what happens when we increase the 

dilation factor to two. For the 

first entry in the feature map, the convolution looks like Figure 

7-8

.

Feature Map

Figure 7-7.  A standard convolution with a dilation factor of one

Chapter 7   temporal Convolutional networks

---

263

Notice that the spacing between each sampled entry has increased by one across all 

directions. Vertically, horizontally, and diagonally, the sampled entries are all spaced 

apart by one entry. Essentially, this spacing is determined by finding what