Pramit Choudhary

WIP(DNNS layerwise visualization)

1

Will update with more details.

Add support for MDLP for BRL

4

Currently, the discretizer implemented to handle continuous features within BRLC is Quantile-based. This could further be improved by adding support for entropy criterion with the Minimum Description Length Principle (MDLP)...

[Interpretable Model] Add support for Interpretable Decision sets

1

This ticket is to extend the coverage for supporting natively interpretable models(_Rule-based model_). Interpretable Decision sets are similar to Bayesian Rule List(BRL) in-terms of the human understandable IF-ELSE conditional statements....

Update conda with the latest binaries

1

To allow the users the flexibility to install either from pip or conda, its important that we maintain consistency on both package managers.

Reference paper for the same: https://arxiv.org/pdf/1309.6392.pdf This algorithm helps in capturing the variance affect across the range of covariates.

Add support for enabling interpretation by visualizing decision boundaries

2

This issue needs to be broken down. The first implementation is just the basic support. Further improvement could be tacked here: https://github.com/datascienceinc/Skater/issues/259

* Improve on existing implementation * Enable support for handling multi-dimensional data Reference: https://staff.fnwi.uva.nl/m.a.migut/migut2015.pdf

Enable support for computing Shapley scores to quantify feature attribution(positive and negative influence) Reference: - Consistent Individualized Feature Attribution for Tree Ensembles (https://arxiv.org/abs/1802.03888) - A Unified Approach to Interpreting Model...

A statistical approach to trace the most influencing data points from the training set responsible for a given prediction Reference: - Understanding Black-box Predictions via Influence Functions(https://arxiv.org/pdf/1703.04730.pdf)

During post pruning of TreeSurrogates some of the not so important nodes gets rejected if the agreed model's performance is not compromised. Will it be possible to add more insight...