noun
vanishing gradient problem
Technical term in deep learning where gradients become extremely small during backpropagation, hindering training of deep neural networks.
深いネットワークでは勾配消失問題が発生しやすい。
In deep networks, the vanishing gradient problem tends to occur.
ReLU関数は勾配消失問題の緩和に役立つ。
The ReLU function helps mitigate the vanishing gradient problem.
The opposite problem where gradients become excessively large, also hindering training.
Compound of 勾配 (gradient), 消失 (vanishing), and 問題 (problem), directly translating the English term 'vanishing gradient problem'.