transfer learning
[/ˈtrænsfər ˈlɜrnɪŋ/]
noun
aprendizado por transferência
1. A machine learning technique where a model trained on one task is reused or adapted for another related task, leveraging knowledge from the source domain to improve performance in the target domain.
Transfer learning allows developers to use a pre-trained image recognition model as a foundation for identifying different types of animals.
O aprendizado por transferência permite que desenvolvedores usem um modelo pré-treinado de reconhecimento de imagens como base para identificar diferentes tipos de animais.
2. The process of applying features, weights, or knowledge learned from one neural network or machine learning model to solve a different but related problem.
Through transfer learning, the model reduced training time by 80% compared to training from scratch.
Por meio do aprendizado por transferência, o modelo reduziu o tempo de treinamento em 80% em comparação com o treinamento do zero.
Transfer learning is a foundational concept in modern machine learning and artificial intelligence development. In Brazil and Portugal, this term is widely used in tech hubs, AI research centers, and software development companies. The concept reflects a global trend in AI efficiency and cost reduction in tech industries. It's particularly important in Brazilian startups and universities focusing on artificial intelligence research.
Related Idioms & Phrases
standing on the shoulders of giants
leveraging existing knowledge
building upon previous work
Look up more words on Fala2Me
The free English-Portuguese dictionary with real Brazilian accents, NYC slang, conjugator and more
Open Fala2Me →