How Weights in Auto Model ( Deep Learning ) Calculated
Hi All,
I just work with RapidMiner and i'm impressed by the simplicity and the accuracy.
I used Auto Model to work on my thesis, and the best one is Deep Learning. But i want to know how the weights is calculated for Deep Learning model. What kind of library and method it's used ? Thanks before. Well done team. Cheers
I just work with RapidMiner and i'm impressed by the simplicity and the accuracy.
I used Auto Model to work on my thesis, and the best one is Deep Learning. But i want to know how the weights is calculated for Deep Learning model. What kind of library and method it's used ? Thanks before. Well done team. Cheers
Find more posts tagged with
Sort by:
1 - 1 of
11
Sort by:
1 - 1 of
11
Hi,
Thanks for your kind words - we really appreciate those 🙏
Here is a description of how those weights are calculated (for all the models, not just Deep Learning):
„RapidMiner uses the averages of the local importances for the model on the validation set to calculate this. Works for every model, but depends on the specific test set. While this in general gives a good idea about what is important in general for the model, the dependency on the test set can make this a bit off sometimes. These weights are shown for each model in Auto Model.“
This means that we explain the predictions on the test set and those columns which contribute more to those explanations get a higher weight. What makes this fantastic is that it works for every model but generates weights specific for that model.
You can find more details here: https://community.rapidminer.com/discussion/comment/64355#Comment_64355
This algorithm was actually implemented as a result of a discussion on this community, namely this one here: https://community.rapidminer.com/discussion/comment/57860#Comment_57860
I always wanted to write a simple paper on this but somehow never found the time 😬
Hope this helps,
Ingo
Thanks for your kind words - we really appreciate those 🙏
Here is a description of how those weights are calculated (for all the models, not just Deep Learning):
„RapidMiner uses the averages of the local importances for the model on the validation set to calculate this. Works for every model, but depends on the specific test set. While this in general gives a good idea about what is important in general for the model, the dependency on the test set can make this a bit off sometimes. These weights are shown for each model in Auto Model.“
This means that we explain the predictions on the test set and those columns which contribute more to those explanations get a higher weight. What makes this fantastic is that it works for every model but generates weights specific for that model.
You can find more details here: https://community.rapidminer.com/discussion/comment/64355#Comment_64355
This algorithm was actually implemented as a result of a discussion on this community, namely this one here: https://community.rapidminer.com/discussion/comment/57860#Comment_57860
I always wanted to write a simple paper on this but somehow never found the time 😬
Hope this helps,
Ingo
Thanks for your kind words - we really appreciate those 🙏
Here is a description of how those weights are calculated (for all the models, not just Deep Learning):
„RapidMiner uses the averages of the local importances for the model on the validation set to calculate this. Works for every model, but depends on the specific test set. While this in general gives a good idea about what is important in general for the model, the dependency on the test set can make this a bit off sometimes. These weights are shown for each model in Auto Model.“
This means that we explain the predictions on the test set and those columns which contribute more to those explanations get a higher weight. What makes this fantastic is that it works for every model but generates weights specific for that model.
You can find more details here: https://community.rapidminer.com/discussion/comment/64355#Comment_64355
This algorithm was actually implemented as a result of a discussion on this community, namely this one here: https://community.rapidminer.com/discussion/comment/57860#Comment_57860
I always wanted to write a simple paper on this but somehow never found the time 😬
Hope this helps,
Ingo