Adding NearestNeighborContrastiveExplainer

[gganapavarapu]

This PR is to contribute Nearest Neighbor Contrastive Explainer algorithm.

Short description of the algorithm is: Nearest Neighbor Contrastive explanation method is an exemplar based constrastive explanation method which provides feasible or realizable contrastive instances. For a given model, exemplar/representative dataset, and query point, it computes the closest point, within the representative dataset, that has a different prediction compared to the query point (with respect to the model).

This black box explainer algorithm is used in industrial use cases. Our group is contributing it to the public AIX360 library to expand the use cases / applications in public domain and invite further development.

Required dependencies:

• numpy

• pandas

• tensorflow

• scikit-learn

• matplotlib (for notebook)

• [x] algorithm folder is added here

• [x] example notebook is added here

[gganapavarapu]

protodash tests failed in latest build which seem to be a connection issue: https://github.com/Trusted-AI/AIX360/actions/runs/4933026686/jobs/8816609988?pr=175. Rerun will pass. (build from forked repo: https://github.com/gganapavarapu/AIX360/actions/runs/4933026442/jobs/8824595758)

[vijay-arya]

Thanks @gganapavarapu

• there are some typos in the nncontrastive notebook (e.g. randome -> random, etc.). Please check.
• Can you add a link to the nncontrastive notebook under examples here

The above corrections could be made along with the next push planned.