Package net.sourceforge.jabm.learning

A library of algorithms for individual learning.


Interface Summary
ActionSelector An action selection policy for a reinforcement-learning algorithm.
ContinuousLearner A learning algorithm that outputs a continuous signal.
DiscreteLearner A learner that learns a discrete number of different actions.
Learner Classes implementing this interface indicate that they implement a learning algorithm.
MDPLearner Classes implementing this interface implement learning algorithms for Markoff descision processes (MDPs).
MimicryLearner A learner that attempts to adjust its output to match a training signal.
SelfKnowledgable Classes implementing this interface indicate that they know if their output is good enough to be used.
StimuliResponseLearner Classes implementing this interface implement myopic stimuli-response reinformcement learning algorithms.

Class Summary
DumbLearner A learner that chooses the same specified action on every iteration.
DumbRandomLearner A learner that simply plays a random action on each iteration without any learning.
EpsilonGreedyActionSelector An implementation of the epsilon-greedy action selection policy.
NPTRothErevLearner A modification of RothErev to address parameter degeneracy, and modified learning with 0-reward.
QLearner An implementation of the Q-learning algorithm.
RothErevLearner A class implementing the Roth-Erev learning algorithm.
SlidingWindowLearner maintains a sliding window over the trained data series and use the average of data items falling into the window as the output learned.
SoftMaxActionSelector An implementation of the softmax action selection policy.
StatelessQLearner A memory-less version of the Q-Learning algorithm.
WidrowHoffLearner An implementation of the Widrow-Hoff learning algorithm for 1-dimensional training sets.

Package net.sourceforge.jabm.learning Description

A library of algorithms for individual learning.