A comprehensive survey of loss functions and metrics in deep learning
Juan R. Terven, Diana Margarita Córdova Esparza, Julio-Alejandro Romero-González, Alfonso Ramírez-Pedraza, Edgar Arturo Chávez‐Urbiola · 2025
Read A comprehensive survey of loss functions and metrics in deep learning by Juan R. Terven, Diana Margarita Córdova Esparza, Julio-Alejandro Romero-González, Alfonso Ramírez-Pedraza, Edgar Arturo Chávez‐Urbiola (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/openalex-w4409349694.txt.
Description
Juan Terven, Diana Córdova-Esparza, Julio-Alejandro Romero-González, and colleagues survey loss functions and metrics in deep learning. Their survey appeared in Artificial Intelligence Review in 2025. Every neural network trains by minimising a loss function. Choosing the wrong one quietly caps a model's quality.
The authors explain how loss functions differ from performance metrics. They set out the properties a good loss should have and note software implementation issues. The survey then works through families of losses and metrics, starting with regression. It is a reference for practitioners who must pick measures for a new task. Clear guidance here can save wasted training runs and misleading evaluations.
Catalog record
| CREATOR | Juan R. Terven, Diana Margarita Córdova Esparza, Julio-Alejandro Romero-González, Alfonso Ramírez-Pedraza, Edgar Arturo Chávez‐Urbiola |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 453 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | OpenAlex / Artificial Intelligence Review |