Contrastive Divergence (CD)
التباعد التبايني
خوارزمية تقريبية لتدريب آلات بولتزمان المقيَّدة: تعرض البيانات على الشبكة، تعيد بناءها من الحالة الخفية، ثم تُعدّل الأوزان لتقريب إعادة البناء من الأصل.
An approximate training algorithm for RBMs that updates weights by comparing original data with its reconstruction.
Also translated asالاختلاف التبايني، CD
First appears in this corpus in: A Learning Algorithm for Boltzmann Machines (1985)
Appears in these papers
- A Learning Algorithm for Boltzmann Machines1985in the sky ✦
- Reducing the Dimensionality of Data with Neural Networks2006in the sky ✦
- A Fast Learning Algorithm for Deep Belief Nets2006in the sky ✦
- Rectified Linear Units Improve Restricted Boltzmann Machines2010in the sky ✦
- Rectified Linear Units Improve Restricted Boltzmann Machines2010in the sky ✦
- Representation Learning: A Review and New Perspectives2013in the sky ✦
- Representation Learning: A Review and New Perspectives2013in the sky ✦