# A Comparative Study of Deep Learning Loss Functions for Multi-Label Remote Sensing Image Classification

arXiv: https://arxiv.org/abs/2009.13935 · HTML: https://mlsanity.com/paper/2009.13935/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Sep 2020.

## Built on by (most cited first, 8 of 8)

- [Evaluating explainable artificial intelligence methods for multi-label deep learning classification tasks in remote sensing](https://mlsanity.com/paper/2104.01375/) (2021)
- [Combining Varied Learners for Binary Classification using Stacked Generalization](https://mlsanity.com/paper/2202.08910/) (2022)
- [Exploiting Memory-aware Q-distribution Prediction for Nuclear Fusion via Modern Hopfield Network](https://mlsanity.com/paper/2410.08889/) (2024)
- [On the Effectiveness of Methods and Metrics for Explainable AI in Remote Sensing Image Scene Classification](https://mlsanity.com/paper/2507.05916/) (2025)
- [A Consensual Collaborative Learning Method for Remote Sensing Image Classification Under Noisy Multi-Labels](https://mlsanity.com/paper/2105.05496/) (2021)
- [Evaluating the Impact of Loss Function Variation in Deep Learning for Classification](https://mlsanity.com/paper/2210.16003/) (2022)
- [Xtreme Margin: A Tunable Loss Function for Binary Classification Problems](https://mlsanity.com/paper/2211.00176/) (2022)
- [A Label Propagation Strategy for CutMix in Multi-Label Remote Sensing Image Classification](https://mlsanity.com/paper/2405.13451/) (2024)

## Abstract (arXiv)

This paper analyzes and compares different deep learning loss functions in the framework of multi-label remote sensing (RS) image scene classification problems. We consider seven loss functions: 1) cross-entropy loss; 2) focal loss; 3) weighted cross-entropy loss; 4) Hamming loss; 5) Huber loss; 6) ranking loss; and 7) sparseMax loss. All the considered loss functions are analyzed for the first time in RS. After a theoretical analysis, an experimental analysis is carried out to compare the considered loss functions in terms of their: 1) overall accuracy; 2) class imbalance awareness (for which the number of samples associated to each class significantly varies); 3) convexibility and differentiability; and 4) learning efficiency (i.e., convergence speed). On the basis of our analysis, some guidelines are derived for a proper selection of a loss function in multi-label RS scene classification problems.