New code example on http://keras.io : classification of timeseries with a 1D convnet, created by Hassan Ismail Fawaz. Uses the FordA dataset of timeseries of engine noise measurements.https://keras.io/examples/timeseries/timeseries_classification_from_scratch/ …
-
Thanks. Isn’t Conv1D not a good fit for timeseries forecasting because in a timeseries, order matters (the recent past is more important than the distant past). Convolution is for translation-invariant problems ?
This is correct, but here we're doing classification, not forecasting. Convolution is a good fit for timeseries classification, in part thanks to translation invariance (it can pick up characteristic patterns independently of their temporal position).
-
-
Thanks for the clarification !
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.