Geometric Origins of Bias in Deep Neural Networks: A Human Visual System Perspective

TL;DR

This paper introduces a geometric analysis framework inspired by the human visual system to understand bias formation in deep neural networks, linking geometric complexity of perceptual manifolds to recognition bias.

Contribution

It presents a novel geometric perspective on bias in DNNs and introduces the Perceptual-Manifold-Geometry library for analyzing manifold properties.

Findings

Differences in geometric complexity lead to recognition biases across categories.

The Perceptual-Manifold-Geometry library has been widely adopted, with over 4,500 downloads.

The work provides a theoretical foundation for developing more equitable AI systems.

Abstract

Bias formation in deep neural networks (DNNs) remains a critical yet poorly understood challenge, influencing both fairness and reliability in artificial intelligence systems. Inspired by the human visual system, which decouples object manifolds through hierarchical processing to achieve object recognition, we propose a geometric analysis framework linking the geometric complexity of class-specific perceptual manifolds in DNNs to model bias. Our findings reveal that differences in geometric complexity can lead to varying recognition capabilities across categories, introducing biases. To support this analysis, we present the Perceptual-Manifold-Geometry library, designed for calculating the geometric properties of perceptual manifolds. The toolkit has been downloaded and installed over 4,500 times. This work provides a novel geometric perspective on bias formation in modern learning systems and lays a theoretical foundation for developing more equitable and robust artificial intelligence.