English

Evaluating the Impact of Loss Function Variation in Deep Learning for Classification

Machine Learning 2022-10-31 v1

Abstract

The loss function is arguably among the most important hyperparameters for a neural network. Many loss functions have been designed to date, making a correct choice nontrivial. However, elaborate justifications regarding the choice of the loss function are not made in related work. This is, as we see it, an indication of a dogmatic mindset in the deep learning community which lacks empirical foundation. In this work, we consider deep neural networks in a supervised classification setting and analyze the impact the choice of loss function has onto the training result. While certain loss functions perform suboptimally, our work empirically shows that under-represented losses such as the KL Divergence can outperform the State-of-the-Art choices significantly, highlighting the need to include the loss function as a tuned hyperparameter rather than a fixed choice.

Keywords

Cite

@article{arxiv.2210.16003,
  title  = {Evaluating the Impact of Loss Function Variation in Deep Learning for Classification},
  author = {Simon Dräger and Jannik Dunkelau},
  journal= {arXiv preprint arXiv:2210.16003},
  year   = {2022}
}
R2 v1 2026-06-28T04:42:24.480Z