Policies.GenericAggregation module

The GenericAggregation aggregation bandit algorithm: use a bandit policy A (master), managing several “slave” algorithms, \(A_1, ..., A_N\).

• At every step, one slave algorithm A_i is selected, by the master policy A.

• Then its decision is listen to, played by the master algorithm, and a feedback reward is received.

• All slaves receive the observation (arm, reward).

• The master also receives the same observation.

Policies.GenericAggregation.random() → x in the interval [0, 1).

class Policies.GenericAggregation.GenericAggregation(nbArms, master=None, children=None, lower=0.0, amplitude=1.0)

Bases: Policies.BasePolicy.BasePolicy

The GenericAggregation aggregation bandit algorithm.

__init__(nbArms, master=None, children=None, lower=0.0, amplitude=1.0)

New policy.

nbArms = None

Number of arms.

lower = None

Lower values for rewards.

amplitude = None

Larger values for rewards.

last_choice = None

Remember the index of the last child trusted for a decision.

children = None

List of slave algorithms.

__str__()

Nicely print the name of the algorithm with its relevant parameters.

startGame()

Start the game for each child, and for the master.

getReward(arm, reward)

Give reward for each child, and for the master.

choice()

Trust one of the slave and listen to his choice.

choiceWithRank(rank=1)

Trust one of the slave and listen to his choiceWithRank.

choiceFromSubSet(availableArms='all')

Trust one of the slave and listen to his choiceFromSubSet.

choiceMultiple(nb=1)

Trust one of the slave and listen to his choiceMultiple.

__module__ = 'Policies.GenericAggregation'

choiceIMP(nb=1, startWithChoiceMultiple=True)

Trust one of the slave and listen to his choiceIMP.

estimatedOrder()

Trust one of the slave and listen to his estimatedOrder.

• Return the estimate order of the arms, as a permutation on \([0,...,K-1]\) that would order the arms by increasing means.

estimatedBestArms(M=1)

Return a (non-necessarily sorted) list of the indexes of the M-best arms. Identify the set M-best.