Value Network
شبكة القيمة
شبكة عصبية تأخذ وضعية اللوح كمدخل وتُخرج قيمة عددية واحدة تمثّل احتمال الفوز من تلك الوضعية، فتحلّ محل آلاف المحاكاة العشوائية بتمريرة أمامية واحدة.
A neural network that takes a board position as input and outputs a single scalar representing the probability of winning from that position, replacing thousands of random simulations with one forward pass.
Also translated asشبكة تقييم الوضعيات، شبكة تقدير الفوز
First appears in this corpus in: Mastering the Game of Go with Deep Neural Networks and Tree Search (2016)
Appears in these papers
- Mastering the Game of Go Without Human Knowledge2017in the sky ✦
- Mastering the Game of Go Without Human Knowledge2017in the sky ✦
- Mastering the Game of Go with Deep Neural Networks and Tree Search2016in the sky ✦
- MuZero: Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model2020in the sky ✦