You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Whenever the user views the whitened / predicted / residual version of the loaded data for the first time, the whitening matrix is calculated and then cached. In the current implementation this calculation is slow to the point that for large datasets, it could take on the order of tens of seconds to calculate the matrix. This happens because the whitening matrix is calculated using the entire dataset.
This is unnecessary since the on-the-fly calculation can be more approximate and use, as is the case in the matlab version, ~5 batches of the data. This would speed up the calculation irrespective of the data size.
Note: the whitening step that's a part of the preprocessing will still use the entire dataset.
The text was updated successfully, but these errors were encountered:
Whenever the user views the whitened / predicted / residual version of the loaded data for the first time, the whitening matrix is calculated and then cached. In the current implementation this calculation is slow to the point that for large datasets, it could take on the order of tens of seconds to calculate the matrix. This happens because the whitening matrix is calculated using the entire dataset.
This is unnecessary since the on-the-fly calculation can be more approximate and use, as is the case in the matlab version, ~5 batches of the data. This would speed up the calculation irrespective of the data size.
Note: the whitening step that's a part of the preprocessing will still use the entire dataset.
The text was updated successfully, but these errors were encountered: