In order to develop machine learning and deep learning models that take into account the guidelines and principles of trustworthy AI, a novel information theoretic approach is introduced in this article. A unified approach to privacy-preserving interpretable and transferable learning is considered for studying and optimizing the trade-offs between the privacy, interpretability, and transferability aspects of trustworthy AI. A variational membership-mapping Bayesian model is used for the analytical approximation of the defined information theoretic measures for privacy leakage, interpretability, and transferability. The approach consists of approximating the information theoretic measures by maximizing a lower-bound using variational optimization. The approach is demonstrated through numerous experiments on benchmark datasets and a real-world biomedical application concerned with the detection of mental stress in individuals using heart rate variability analysis.
A. Geyer-Schulz. Genetic Algorithms and Soft Computing, volume 8 of Studies in Fuzziness and Soft Computing, page 658--682. Heidelberg, Physica-Verlag, (September 1996)
J. Baldwin, T. Martin, and J. Shanahan. Fuzzy Logic in Artificial Intelligence, IJCAI'97
Workshop, Selected and Invited Papers, volume 1566 of Lecture Notes in Artificial Intelligence, page 91--116. Nagoya, Japan, Springer, (August 1997)Published 1999.
M. Akbarzadeh-T., E. Tunstel, K. Kumbla, and M. Jamshidi. Proceedings of the 1998 IEEE World Congress on
Computational Intelligence, 2, page 1200--1205. Anchorage, Alaska, USA, IEEE Press, (5-9 May 1998)