Backpropagation
[/ˌbækprɒpəˈɡeɪʃən/]
noun
retropropagação
1. A machine learning algorithm used to train artificial neural networks by calculating the gradient of the loss function with respect to each weight by the chain rule, propagating the error backwards through the network layers
Backpropagation is the fundamental algorithm that enables deep learning models to learn from data by adjusting weights based on prediction errors.
A retropropagação é o algoritmo fundamental que permite que modelos de aprendizado profundo aprendam com os dados ajustando os pesos com base nos erros de previsão.
2. The process of computing gradients in a neural network by working backwards from the output layer to the input layer
During backpropagation, the network calculates how much each parameter contributed to the overall error.
Durante a retropropagação, a rede calcula quanto cada parâmetro contribuiu para o erro geral.
Backpropagation is a cornerstone concept in artificial intelligence and machine learning, originating from research in the 1980s-90s. The term is widely used in English in Brazilian academic and technology sectors, though the Portuguese translation 'retropropagação' is employed in formal Portuguese-language literature. In both Brazil and the USA, professionals often use the English term when discussing AI/ML concepts, as the field's primary literature and communication happens in English.
Look up more words on Fala2Me
The free English-Portuguese dictionary with real Brazilian accents, NYC slang, conjugator and more
Open Fala2Me →