2026-04-13T00:11:06Zhttps://recercat.cat/oai/request

oai:recercat.cat:2117/3739032026-01-21T06:09:45Zcom_2072_1033col_2072_452950

Focus! Rating XAI methods and finding biases Arias Duart, Anna Parés Pont, Ferran Garcia Gasulla, Dario Giménez Ábalos, Víctor Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial::Aprenentatge automàtic Deep learning Pattern recognition systems Neural networks (Computer science) Measurement Visualization Computational modeling Computer architecture Robustness Behavioral sciences Labeling Aprenentatge profund Reconeixement de formes (Informàtica) Xarxes neuronals (Informàtica) AI explainability improves the transparency and trustworthiness of models. However, in the domain of images, where deep learning has succeeded the most, explainability is still poorly assessed. In the field of image recognition many feature attribution methods have been proposed with the purpose of explaining a model’s behavior using visual cues. However, no metrics have been established so far to assess and select these methods objectively. In this paper we propose a consistent evaluation score for feature attribution methods—the Focus—designed to quantify their coherency to the task. While most previous work adds outof-distribution noise to samples, we introduce a methodology to add noise from within the distribution. This is done through mosaics of instances from different classes, and the explanations these generate. On those, we compute a visual pseudo-precision metric, Focus. First, we show the robustness of the approach through a set of randomization experiments. Then we use Focus to compare six popular explainability techniques across several CNN architectures and classification datasets. Our results find some methods to be consistently reliable (LRP, GradCAM), while others produce class-agnostic explanations (SmoothGrad, IG). Finally we introduce another application of Focus, using it for the identification and characterization of biases found in models. This empowers bias-management tools, in another small step towards trustworthy AI. This work is supported by the European Union – H2020 Program under the “INFRAIA-01-2018-2019 – Integrating Activities for Advanced Communities”, Grant Agreement n.871042, “SoBigData++: European Integrated Infrastructure for Social Mining and Big Data Analytics” and by the Dept. de Recerca i Universitats of the Generalitat de Catalunya under the Industrial Doctorate Grant DI 2018-100. Peer Reviewed Postprint (author's final draft) 2022 Conference report https://ieeexplore.ieee.org/document/9882821 info:eu-repo/grantAgreement/EC/H2020/871042/EU/SoBigData++: European Integrated Infrastructure for Social Mining and Big Data Analytics/SoBigData-PlusPlus Open Access Institute of Electrical and Electronics Engineers (IEEE)