parameterized greedy policy

A reinforcement learning strategy that selects actions based on current best estimates while maintaining the flexibility to adapt over time through parameters.