What did AlphaGo's policy network predict?
It assigned probabilities to possible moves from a board position. That guidance helped the system focus its search on moves considered more promising.
Explore This Question
Search for additional explanations using these tools:
More FAQs in How did AlphaGo use supervised learning and reinforcement learning?
Move selection and position evaluation are different tasks. A value estimate helps assess how favorable a position is without fully exploring every possible continuation.
No. The original system used expert games in supervised training. The later AlphaGo Zero system removed that dependency and learned from self-play instead.
Still have questions?
Browse all our FAQs or reach out to our support team