ModifiedPolicyIterationAgent<T>
Models & Types ยท AiDotNet.ReinforcementLearning.Agents.DynamicProgramming
Modified Policy Iteration agent - hybrid of Policy Iteration and Value Iteration.
Inheritance: ReinforcementLearningAgentBase<T> โ ModifiedPolicyIterationAgent<T>
For Beginners
Modified Policy Iteration is a middle ground between two classic algorithms: Value Iteration (fast but less stable) and Policy Iteration (stable but slow). Instead of fully evaluating a policy before improving it, it does a limited number of evaluation sweeps. Think of it like proofreading a draft: you do a few passes (not infinite) before revising. The number of evaluation sweeps controls the speed-stability trade-off.
How It Works
Modified PI performs limited policy evaluation sweeps before improvement, trading off between the efficiency of VI and the stability of PI.
Example
// Create a Modified Policy Iteration agent balancing speed and stability
var options = new ModifiedPolicyIterationOptions<double> { StateSize = 4, ActionSize = 2, MaxEvaluationSweeps = 5 };
var agent = new ModifiedPolicyIterationAgent<double>(options);
// Select an action using the partially-evaluated policy
var state = new Vector<double>(new double[] { 0.5, -0.3, 1.0, 0.2 });
var action = agent.SelectAction(state);
Constructors
| Constructor | Summary |
|---|---|
ModifiedPolicyIterationAgent() |
Initializes a new instance with default options (StateSize=4, ActionSize=2). |
ModifiedPolicyIterationAgent(ModifiedPolicyIterationOptions<T> options) |
Properties
| Property | Summary |
|---|---|
int FeatureCount { get; } |
Methods
| Method | Summary |
|---|---|
Dictionary<string, T> GetMetrics() |
|
ModelMetadata<T> GetModelMetadata() |
|
T Train() |
|
Task TrainAsync() |
|
Task<Vector<T>> PredictAsync(Vector<T> input) |
|
Vector<T> Predict(Vector<T> input) |
|
Vector<T> SelectAction(Vector<T> state, bool training) |
|
byte[] Serialize() |
Auto-generated common model serialization surface. |
void Deserialize(byte[] data) |
Auto-generated common model deserialization surface. |
void LoadModel(string filepath) |
|
void ResetEpisode() |
|
void SaveModel(string filepath) |
|
void StoreExperience(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done) |