A Fair Classifier Embracing Triplet Collapse
arXiv:2306.04400v1h-index: 39
Originality Synthesis-oriented
AI Analysis
This addresses fairness issues in machine learning, but appears incremental as it builds on existing triplet loss methods.
The paper tackles the problem of bias in machine learning models by exploiting the collapse of the triplet loss under specific conditions, resulting in a fair classifier that limits biases.
In this paper, we study the behaviour of the triplet loss and show that it can be exploited to limit the biases created and perpetuated by machine learning models. Our fair classifier uses the collapse of the triplet loss when its margin is greater than the maximum distance between two points in the latent space, in the case of stochastic triplet selection.