Some Information Geometric Aspects of Cyber Security by Face Recognition

Area of Science:

• Computer Science
• Information Geometry
• Biometrics

Background:

• Secure user access relies on biometrics like fingerprint and face recognition.
• Managing large datasets of secure digital objects requires efficient classification and feature retrieval.
• Multidimensional methods are needed for data represented by probability distributions.

Purpose of the Study:

• To propose a new face recognition method leveraging information geometry.
• To utilize geodesic distances as a dissimilarity metric for image comparison.
• To enhance recognition rates in secure user access systems.

Main Methods:

• Representing probability distributions as Riemannian manifolds.
• Calculating dissimilarities using geodesic distances between distributions.
• Developing a face recognition algorithm based on RGB Gaussian approximations and joint probabilities.

Main Results:

• The proposed method effectively identifies clusters and projections for easier interpretation.
• Geodesic distances provide a natural metric for comparing data described by probability distributions.
• Experimental results demonstrate superior recognition rates compared to state-of-the-art methods.

Conclusions:

• Information geometry offers a robust framework for analyzing complex data distributions.
• The novel face recognition method shows significant promise for secure access applications.
• This approach advances the field of biometrics through advanced geometric principles.