label noise
[/ˈleɪbəl nɔɪz/]
noun
ruído de rótulo
1. In machine learning, errors or inaccuracies in the labeled training data where the assigned class or target value is incorrect or misclassified
The model's performance was affected by label noise in the training dataset where some images were incorrectly annotated.
O desempenho do modelo foi afetado pelo ruído de rótulo no conjunto de dados de treinamento, onde algumas imagens foram anotadas incorretamente.
2. Corruption or mislabeling of data that introduces errors into supervised learning algorithms
Researchers developed techniques to handle label noise and improve model robustness.
Pesquisadores desenvolveram técnicas para lidar com ruído de rótulo e melhorar a robustez do modelo.
This is a specialized technical term primarily used in machine learning and data science communities. It is most common in academic research, AI development, and data annotation discussions. The term reflects the growing importance of data quality in AI/ML development, particularly relevant as Brazilian and Portuguese tech sectors expand in AI research.
Look up more words on Fala2Me
The free English-Portuguese dictionary with real Brazilian accents, NYC slang, conjugator and more
Open Fala2Me →