The target parameter of the AE can accept 'full', 'channels', 'timeseries'. Timeseries is not implemented but channels uses an autoencoder that seems to be able to handle either channels or timeseries (as it takes output_dim AND output_dim_2).
What's the difference between what the single and double AE can handle? And can the single AE work for either channels or timeseries? In other words, should we change the target parameter and only have two options?
The target parameter of the AE can accept 'full', 'channels', 'timeseries'. Timeseries is not implemented but channels uses an autoencoder that seems to be able to handle either channels or timeseries (as it takes output_dim AND output_dim_2).
What's the difference between what the single and double AE can handle? And can the single AE work for either channels or timeseries? In other words, should we change the target parameter and only have two options?