Class PrioritizedReplayBuffer<T>

Namespace

AiDotNet.ReinforcementLearning.ReplayBuffers

Assembly

AiDotNet.dll

Prioritized experience replay buffer for reinforcement learning.

public class PrioritizedReplayBuffer<T>

Type Parameters

T

The numeric type used for calculations.

Inheritance

object

PrioritizedReplayBuffer<T>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Prioritized replay samples important experiences more frequently based on TD error. Uses sum tree data structure for efficient sampling.

Constructors

PrioritizedReplayBuffer(int)

public PrioritizedReplayBuffer(int capacity)

Parameters

capacity int

Properties

Count

public int Count { get; }

Property Value

int

Methods

Add(Vector<T>, Vector<T>, T, Vector<T>, bool)

public void Add(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done)

Parameters

state Vector<T>

action Vector<T>

reward T

nextState Vector<T>

done bool

Sample(int, double, double)

public (List<(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done)> batch, List<int> indices, List<double> weights) Sample(int batchSize, double alpha, double beta)

Parameters

batchSize int

alpha double

beta double

Returns

(List<(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done)> batch, List<int> indices, List<double> weights)

UpdatePriorities(List<int>, List<double>, double)

public void UpdatePriorities(List<int> indices, List<double> priorities, double epsilon)

Parameters

indices List<int>

priorities List<double>

epsilon double