Package | Description |
---|---|
net.sourceforge.jabm.learning |
A library of algorithms for individual learning.
|
Modifier and Type | Interface and Description |
---|---|
interface |
ContinuousLearner
A learning algorithm that outputs a continuous signal.
|
interface |
DiscreteLearner
A learner that learns a discrete number of different actions.
|
interface |
MDPLearner
Classes implementing this interface implement learning algorithms for Markoff
descision processes (MDPs).
|
interface |
MimicryLearner
A learner that attempts to adjust its output to match a training signal.
|
interface |
StimuliResponseLearner
Classes implementing this interface implement myopic stimuli-response
reinformcement learning algorithms.
|
Modifier and Type | Class and Description |
---|---|
class |
AbstractLearner |
class |
DumbLearner
A learner that chooses the same specified action on every iteration.
|
class |
DumbRandomLearner
A learner that simply plays a random action on each iteration without any
learning.
|
class |
MetaLearner |
class |
NPTRothErevLearner
A modification of RothErev to address parameter degeneracy, and modified
learning with 0-reward.
|
class |
QLearner
An implementation of the Q-learning algorithm.
|
class |
RothErevLearner
A class implementing the Roth-Erev learning algorithm.
|
class |
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.
|
class |
StatelessQLearner
A memory-less version of the Q-Learning algorithm.
|
class |
WidrowHoffLearner
An implementation of the Widrow-Hoff learning algorithm for 1-dimensional
training sets.
|
class |
WidrowHoffLearnerWithMomentum |
Copyright © 2014. All rights reserved.