1
$\begingroup$

guys.

I have been learning about machine learning using Azure ML service from Microsoft and my team and I have seen that there are some multi-class decision algorithms, the Forest and Jungle are examples.

We don't know what's the real difference between these ones (we had checked the documentation of them in https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/multiclass-decision-forest and https://docs.microsoft.com/en-us/azure/machine-learning/studio-module-reference/multiclass-decision-jungle), because they look very similar.

Could you help us with that? What's the difference between these algorithms?

Thank you.

$\endgroup$
1
$\begingroup$

The link you provided on Decision Jungles states that they split the space by building directed acyclig graphs. Random Forests use trees, which are only a special case of DAGs. It seems like the algorithm is able to fuse nodes, something that isn't happening in constructing a decision tree.

If you follow the link on Decision Jungle's implementation you'll find a paper that explains the method.

From the paper, figure 1:

Using DAGs instead of trees reduces the number of nodes and can result in better generalization.

For a concrete visualization see page 4 of the article supplement - it clearly shows what is built is not a tree, and something more complex.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.