Can anyone give me a better understanding on how convultions would work on sequence data?
Good morning, Laura!
That’s a great question. In images, convolutions are used to identify specific patterns in small regions, such as an ear, a tail, or a nose. The concept is similar in time series or sequential data: a filter is applied to several windows of data to extract local characteristics, such as trends or cycles.
Please let me know if you have any further questions.