Dataset of Train Latent ODE Network with Irregularly Sampled Time-Series Data

hello,

I'am working for re-construct the code of

Train Latent ODE Network with Irregularly Sampled Time-Series Data

https://www.mathworks.com/help/deeplearning/ug/train-latent-ode-network-with-irregularly-sampled-time-series-data.html

but I stuck in understanding structure of dataset ("irregularSineWaves" dataset (1x1000x100)).

there are "squeeze" function to removes dimensions of an array that have size 1 (changes to 1000x100).

my question is that this code train first irregularsinewaves ??? (:,1), or train averaged 100 irregularsinewaves??

The structure of my dataset is only value and time (normal timeseries data).

If anyone has applied this code on normal timeseries dataset, please help me how to fix the code.

Answer 1

[Conor Daly]

The irregular sine waves dataset consists of 1000 time series, each with 100 (irregularly sampled) time steps.

The example https://www.mathworks.com/help/deeplearning/ug/train-latent-ode-network-with-irregularly-sampled-time-series-data.html trains a network on all 800 time series (an 8:2 train/test split is used), by training on batches of 50 per iteration.

It sounds like your data is a single time series.

[minhyuk jeung]

Thank you.