Attribution-based confidence score - Classification

Confidence score computation based on Neighborhood sampling for image classification.

This method has been successfully used for image classification as done in [Jha et al., 2019] with attribution-based confidence (ABC). In contrast with other approaches, it does not need a meta-model for supervision nor does it require access to training data, making it compatible with black-box and privacy sensitive scenarios. The core of the ABC algorithm is focused on improving performance by sampling mainly around pixels that have a high influence on the output.

Documentation

Methodological Guidelines

• Methodological guideline for Robustness Functional Set
• Methodological Guideline and Evaluation tools for robust artificial intelligence

Benchmark

• Report on Confidence Score improvement methods

State of the art

• Characterization of the notion of trust
• State of the Art on Data based AI Confidence Evaluation

Demonstrator

Associated Demonstrator relies on Visual Inspection use-case.

Support

Support for Attribution-based confidence score - Classification must be obtained by sending an email to support@confiance.ai

Ensure your email contains :

• Your name
• A link to this page
• the version you are working with
• A clear description of the problematic (bug, crash, feature or help request)
• A full description of the problem whichallow to reproduce it
• Any file or screenshort element mandatory for the full understanding of the problematic