Model-Guided Intelligent Search
Definition
The core algorithmic approach used by all DeepMind's Alpha systems. The system builds a learned model of the problem space based on training data, then uses that model to guide an intelligent search (Monte Carlo tree search, reinforcement learning, etc.) through the enormous combinatorial space to find solutions. This is more efficient than brute-force search because the model identifies promising directions.
Cast within
No topic-region cast yet — this appears once Model-Guided Intelligent Search's compiled claims are aligned into a topic region's argument tree.
Claims mentioning Model-Guided Intelligent Search (1)
Your notes
Loading notes...