For example, let say for each time period there are four variables and the prediction output is not one of the variables but the same 4 variables.
tw1: [[1,2,3,4]
tw2: [2,3,4,5]
tw3: [3,4,5,6]
...
If the input size is 2, with trainX sample of [1,2,3,4] [2,3,4,5],
the train y sample is [3,4,5,6]
Hi, thank you very much for the package!
I have checked the notebook of using Lime on LSTM [here].(https://github.com/marcotcr/lime/blob/master/doc/notebooks/Lime%20with%20Recurrent%20Neural%20Networks.ipynb) I wonder if it can be still used for multi parallel output LSTM prediction.
For example, let say for each time period there are four variables and the prediction output is not one of the variables but the same 4 variables.
tw1: [[1,2,3,4] tw2: [2,3,4,5] tw3: [3,4,5,6] ... If the input size is 2, with trainX sample of [1,2,3,4] [2,3,4,5], the train y sample is [3,4,5,6]
Can we explain these cases with Lime as well?