[01/08/2019] Version 0.0.1 released!

Today we release our first public version of the FAT Forensics package. To familiarise yourself with the FAT Forensics Python package make sure to check out our Getting Started page. The changelog summarises functionality of this release: 0.0.1 (01/08/2019).

[04/11/2019] Version 0.0.2 released!

Today we release an incremental update focused on surrogate explainability of black-box models for tabular data – a collection of techniques and algorithms that we call build LIME yourself (bLIMEy). The changelog summarises the functionality added with this release: 0.0.2 (04/11/2019).

This code release comes with a new how-to guide: How to build LIME yourself (bLIMEy) – Surrogate Tabular Explainers. We have also added one more code example – Using a Surrogate Tree Explainer – and updated the LIME code example (Using LIME Explainer) to use the bLIMEy implementation of LIME (fatf.transparency.predictions.surrogate_explainers.TabularBlimeyLime) instead. The “Explaining Machine Learning Predictions: LIME and Counterfactuals” tutorial has also been updated to use bLIMEy instead of LIME.

Surrogate explainability for image and text data is coming soon so stay tuned.