Skip to content

Calibrate to Interpret

Trustworthy machine learning is driving a large number of ML community works in order to improve ML acceptance and adoption. The main aspect of trustworthy machine learning are the followings: fairness, uncertainty, robustness, explainability and formal guaranties. Each of these individual domains gains the ML community interest, visible by the number of related publications. However, few works tackle the interconnection between these fields. In this paper, we show a first link between uncertainty and explainability, by studying the relation between calibration and interpretation. As the calibration of a given model changes the way it scores samples, and interpretation approaches often rely on these scores, it seems safe to assume that the confidence-calibration of a model interacts with our ability to interpret such model. In this paper, we show, in the context of networks trained on image classification tasks, to what extent interpretations are sensitive to confidence-calibration. It leads us to suggest a simple practice to improve the interpretation outcomes : Calibrate to Interpret.

Gregory Scafarto, Nicolas Posocco, Antoine Bonnefoy, Calibrate to Interpret, In Proc. of The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2022), September 2022.

Click here to access the paper.

Releated Posts

Investigating a Feature Unlearning Bias Mitigation Technique for Cancer-type Bias in AutoPet Dataset

We proposed a feature unlearning technique to reduce cancer-type bias, which improved segmentation accuracy while promoting fairness across sub-groups, even with limited data.
Read More

Muppet: A Modular and Constructive Decomposition for Perturbation-based Explanation Methods

The topic of explainable AI has recently received attention driven by a growing awareness of the need for transparent and accountable AI. In this paper, we propose a novel methodology to decompose any state-of-the-art perturbation-based explainability approach into four blocks. In addition, we provide Muppet: an open-source Python library for explainable AI.
Read More