This message was deleted Nixtla Community #neural-forecast

Join Slack

This message was deleted.

# neural-forecast

Slackbot

09/22/2023, 3:14 PM

This message was deleted.

Cristian (Nixtla)

09/25/2023, 7:49 PM

Hi @Phil! You understanding on both parameters is correct. Regarding the inverse, not necessarily. The

n_pool_kernel_size

downsamples only the inputs, and the

n_freq_downsample

only the output. Having both at the same time do not compound, because they affect different parts of the architecture. Here is the diagram of the paper, the kernel controls the Maxpool im the inputs of the MLP stack, and

n_freq_downsample

controls the output dimension of theta (points in the forecasting window)

Cristian (Nixtla)

09/25/2023, 7:50 PM

The intuition of having both larger at the same time is that to output a lower dimensional output (higher

n_freq

) you need less information from the inputs, so kernel is larger

Cristian (Nixtla)

09/25/2023, 7:52 PM

With that said, we have observed that larger kernel sizes only help in very high frequency data, and usually keeping a value of 1 or 2 is the best. That is why in our default config we kept the option of no downsampling ([1,1,1])

Phil

09/25/2023, 7:54 PM

I see that makes more sense! Thank you! Have you observed any effects of varying the number of blocks across frequencies. or keeping them constant is roughly equivalent in performance

Cristian (Nixtla)

09/25/2023, 9:56 PM

We recommend increasing the blocks with larger datasets. For example, the NBEATS uses 30 blocks in total for each frequency of the M4 dataset, with around 30k series

✅ 1

6 Views

Open in Slack

Previous Next