|
gptkbp:instance_of
|
gptkb:Artificial_Intelligence
gptkb:program
|
|
gptkbp:aifield_impact
|
gptkb:significant
|
|
gptkbp:architecture
|
gptkb:neural_networks
|
|
gptkbp:coat_of_arms
|
gptkb:20
|
|
gptkbp:community_feedback
|
highly positive
|
|
gptkbp:contributed_to
|
machine learning advancements
|
|
gptkbp:developed_by
|
gptkb:Deep_Mind
|
|
gptkbp:difficulty
|
high
|
|
gptkbp:evaluates
|
gptkb:Monte_Carlo_Tree_Search
win rate
self-play games
|
|
gptkbp:field_of_study
|
gptkb:Artificial_Intelligence
|
|
gptkbp:final_match_date
|
won
|
|
gptkbp:first_match
|
Alpha Go Lee
|
|
gptkbp:future_prospects
|
AI applications in various fields
|
|
gptkbp:game_type
|
gptkb:Go
|
|
gptkbp:has_achievements
|
multiple tournament victories
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Alpha Go Zero
|
|
gptkbp:impact_on_go
|
revolutionized strategies
|
|
gptkbp:influenced
|
AI game playing
|
|
gptkbp:input_output
|
move probabilities
|
|
gptkbp:input_representation
|
board position
|
|
gptkbp:inspired
|
further AI developments
|
|
gptkbp:is_compared_to
|
gptkb:Alpha_Go_Master
|
|
gptkbp:is_taught_in
|
adaptive
|
|
gptkbp:legacy
|
pioneering AI in games
|
|
gptkbp:no_human_data
|
gptkb:true
|
|
gptkbp:notable_feature
|
no prior knowledge
learns without human data
|
|
gptkbp:number_of_games
|
over 100,000
|
|
gptkbp:orbital_period
|
40 million
|
|
gptkbp:performance
|
exceeded human players
|
|
gptkbp:policy_network
|
used
|
|
gptkbp:provides_information_on
|
generated by itself
|
|
gptkbp:publications
|
multiple research papers
|
|
gptkbp:rank
|
superhuman
top Go player
|
|
gptkbp:release_date
|
October 2017
|
|
gptkbp:release_year
|
gptkb:2017
|
|
gptkbp:result
|
dominant performance
|
|
gptkbp:self_improvement
|
continuous
|
|
gptkbp:self_play
|
gptkb:true
|
|
gptkbp:significance
|
gptkb:AI_technology
No human data used for training
|
|
gptkbp:strategic_importance
|
high
|
|
gptkbp:strategy
|
aggressive and innovative
|
|
gptkbp:successor
|
gptkb:Alpha_Go
|
|
gptkbp:training
|
several days
reinforcement learning
iterative improvement
Self-Play
|
|
gptkbp:training_algorithm
|
gptkb:Alpha_Zero
|
|
gptkbp:training_programs
|
distributed computing
|
|
gptkbp:training_speed
|
gptkb:fast
|
|
gptkbp:user_base
|
researchers and developers
|
|
gptkbp:value_network
|
used
|
|
gptkbp:was_a_demonstration_of
|
generalization ability
|
|
gptkbp:bfsParent
|
gptkb:Deep_Mind
gptkb:Alpha_Go
|
|
gptkbp:bfsLayer
|
5
|