AntakIA is our Python library that offers a first level of regional AI explainability, open-source.
AntakIA runs in a Jupyter notebook or equivalent.
You can install it via pip install antakia
and use it with one of the provided notebooks.
demo
(or demo1
, demo2
, demo3
, demo4
, or demo5
) and the password antakia
.
The documentation for using AntakIA is accessible via the Github repository or through this link. AntakIA is a Python library that supports all stages of our regional approach:
Our regional explanation approach first aims to decompose a complicated problem into simpler elements. We thus divide the space of possible values into regions and then substitute the original model with simpler models within those regions.
Illustratively, like with a mosaic, we break down the space into different parcels, where we place our tiles. It is this allegory of the mosaic that made us think of the ancient city of Antioch, renowned for its Roman mosaics. Today, Antioch is in Turkey, and is called Antakya, which we write as AntakIA, with an IA 🙂
Note that AntakIA is also the acronym for: A New Tool To Acquire Knowledge from AI.