Time Series Classification, Not Improving Beyond First Epoch

1

Hello,

Full disclosure, I am new to PyTorch and machine learning in general.

I am trying to follow this tutorial to preform Time Series Classification. However, instead of using the data set provided in the tutorial, I substituted it with my own data set.

However, when I run the code with my data set it does not seem to improve beyond the very first epoch. Does anyone have any ideas as to what I could experiment with to try and fix this? I’ve tried experimenting with different drop out rates but that did not seem to any have noticeable effect. Perhaps I forgot to make and obvious change somewhere.

You can view the modified code with my data set here,

https://colab.research.google.com/drive/1H048DMQlpSCMn7fh_1HsGa6RhdAcU233?usp=sharing

Thanks!

2

You could use a small subset of the dataset (e.g. just 10 samples) and try to overfit your model on it by playing around with the hyperparameters etc. Once this is done you could scale up the use case again.

3

Hello! I’ve been experimenting with different parameters for the past few days, but I haven’t made any significant progress.

Are there any PyTorch tutorials that show how to do multivariate time series classification?

Thanks!