Problem with table/cell array in an LSTM anomaly detection project

Hi everyone, I'm a novice of MATLAB and programming in general and I'm doing an anomaly detection project through LSTM neural networks.

My problem is at the origin of the project since I can't understand how to treat the dataset I want to use.

Being a .csv file, this is represented, inside MATLAB, with a table, but reading many examples on LSTM, I noticed that all the datasets used were never table but array or cell array.

I would like to ask you if it is possible to use a table in this kind of problem or if the dataset should be transformed first by some function and if so which ones (table2cell?).

I leave you a picture of what the table looks like once opened in MATLAB, in this one you can see a column that represents the day and time of the detection, 52 sensors and a final column that represents the state of the system and that is the target of our problem:

Obviously it is not a 4x9 matrix but 220000x55, reduced only as an example.

Thank you to everyone who would like to help me.

Answer 1

[Divya Gaddipati]

You can use a table as an input to the trainNetwork function for training your LSTM model.

If the data size is same for all columns, then table can be used. Otherwise, it is preferable to use cell.

You can refer to the usage of trainNetwork in the below link:

https://www.mathworks.com/help/deeplearning/ref/trainnetwork.html#mw_2323cdfe-e56b-4f7f-8195-b345e277aad4

Hope this helps!