shapley value

A shapley value is a solution concept from cooperative game theory that distributes a finite pool of surplus among participants in proportion to their average marginal contribution across every possible coalition. It is defined by four axioms: efficiency (the total surplus is fully distributed), symmetry (players contributing equally receive equal shares), the dummy property (a player who adds nothing to any coalition receives zero), and additivity (the value of a combined game equals the sum of values from each subgame). In machine learning it is adapted as SHAP — assigning feature attributions by evaluating a model's output across all possible subsets of features, approximated via Monte Carlo sampling when exact enumeration is infeasible. [formal: valor Shapley | substrate: mind | horizon: a life | explicit: yes | epoch: 0.42]