4 ms·
DNN's make good features though, especially if you have time series data or lots of text. I agree that the final model should be a randomforest/xgboost/lightgb
by psandersen 8y ago
DNN's make good features though, especially if you have time series data or lots of text.
I agree that the final model should be a randomforest/xgboost/lightgbm for typical tabular data.
- mlthoughts2018 8y agoI meant that extracting an intermediate layer as a feature embedding and then sticking a classical model on top of it performs worse than curating features through domain-specific expert tuning, for a ton of diverse application domains.