A Fair Loss Function for Network Pruning
It addresses fairness issues in pruning for resource-constrained deployments, but is incremental as it modifies existing loss functions.
The paper tackles the problem of pruning neural networks exacerbating biases, and introduces a performance weighted loss function that limits bias introduction during pruning, showing effectiveness on datasets like CelebA, Fitzpatrick17k, and CIFAR-10.
Model pruning can enable the deployment of neural networks in environments with resource constraints. While pruning may have a small effect on the overall performance of the model, it can exacerbate existing biases into the model such that subsets of samples see significantly degraded performance. In this paper, we introduce the performance weighted loss function, a simple modified cross-entropy loss function that can be used to limit the introduction of biases during pruning. Experiments using the CelebA, Fitzpatrick17k and CIFAR-10 datasets demonstrate that the proposed method is a simple and effective tool that can enable existing pruning methods to be used in fairness sensitive contexts. Code used to produce all experiments contained in this paper can be found at https://github.com/robbiemeyer/pw_loss_pruning.