2021 IEEE International Conference on Cyber Security and Resilience

Full Program

Summary:

Machine learning has been thriving during the last two decades on the strength of some key factors such as the creation of large datasets through automated procedures.
However, datasets often contain biases that can significantly affect the performance and resilience of machine learning models.
This paper presents X-BaD a tool for bias detection designed to inject and discover biases in a neural network. It is implemented as an open-source Python library that extends the Spectral Relevance Analysis methodology. It allows data reusability and user customization by parameter configurations and offers built-in functions to inject artificial biases into popular image datasets for test purposes. This tool is compatible and extensible with features that are commonly used in machine learning frameworks, such as PyTorch or Sci-kit Learn. It also includes functions to interpret and assess the processed data. Finally, a set of experiments are presented to validate the proposed approach.

Author(s):

Marco Pacini    
Department of Excellence in Robotics & AI, Scuola Superiore Sant'Anna
Italy

Federico Nesti    
Department of Excellence in Robotics & AI, Scuola Superiore Sant'Anna
Italy

Alessandro Biondi    
Department of Excellence in Robotics & AI, Scuola Superiore Sant'Anna
Italy

Giorgio Buttazzo    
Department of Excellence in Robotics & AI, Scuola Superiore Sant'Anna
Italy

 


Copyright © 2021 SUMMIT-TEC GROUP LTD