Additive Attention

الانتباه الجمعي

آلية تقييم انتباه تحسب درجة الصلة بجمع إسقاطين خطيين (لحالة فاكّ الترميز وتمثيل المُرمِّز) ثم تطبيق دالة tanh، بخلاف الانتباه الضربي الذي يستخدم الضرب النقطي.

An attention scoring mechanism that computes relevance by adding two linear projections (of the decoder state and encoder annotation) then applying tanh, in contrast to multiplicative attention which uses a dot product.

Also translated asانتباه باهداناو، الانتباه التسلسلي، الانتباه بالجمع

First appears in this corpus in: Neural Machine Translation by Jointly Learning to Align and Translate (2014)

Appears in these papers

Neural Machine Translation by Jointly Learning to Align and Translate2014 in the sky ✦
Google's Neural Machine Translation System: Bridging the Gap Between Human and Machine Translation2016 in the sky ✦
Recurrent Neural Networks (RNNs): A Gentle Introduction and Overview2019 in the sky ✦
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention2015 in the sky ✦
Show, Attend and Tell: Neural Image Caption Generation with Visual Attention2015 in the sky ✦
Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions2018 in the sky ✦