okrasolar / pytorch-timeseries

PyTorch implementations of neural networks for timeseries classification
107 stars 23 forks source link

InceptionTime blocks incorrectly implemented #2

Open RossDeVito opened 3 years ago

RossDeVito commented 3 years ago

The general flow of data for an inception block is:

input data -> bottleneck -> [filters of different sizes] -> output

In your implementation, the filters of different sizes are applied sequentially like this:

input data -> bottleneck -> filter_size_a -> filter_size_b -> filter_size_c  -> output

In the actual implementation the flow is something like the following.

input data -> bottleneck -> { -> filter_size_a ->         } -> concat-> output
                                -> filter_size_b ->
                                -> filter_size_c ->
                                -> max_pool_filter ->
dvirnimrod commented 3 years ago

I notice that too @RossDeVito and absolutely agree.

jumdc commented 1 year ago

Same here.