FinancialDQNAgent<T>
Models & Types · AiDotNet.Finance.Trading.Agents
Financial Deep Q-Network (DQN) agent for discrete action trading.
Inheritance: ReinforcementLearningAgentBase<T> → TradingAgentBase<T> → FinancialDQNAgent<T>
Implements: IGeneratedParameterRegistrar<T>, IGradientComputable<T, Vector<T>, Vector<T>>, IMaskableAgent<T>
For Beginners
The DQN (Deep Q-Network) trading agent learns to make discrete trading decisions (buy, sell, or hold) by estimating the long-term value of each action. It maintains a “memory” of past experiences and learns from random samples of those memories. A separate target network prevents the learning from becoming unstable. DQN is best suited for trading scenarios with a fixed set of possible actions.
Example
// Define Q-network architecture for discrete trading (10 state features, 3 actions: buy/hold/sell)
var architecture = new NeuralNetworkArchitecture<double>(
inputType: InputType.OneDimensional,
taskType: NeuralNetworkTaskType.Regression,
inputSize: 10, outputSize: 3);
// Create DQN agent for discrete action stock trading
var options = new TradingAgentOptions<double>();
var model = new FinancialDQNAgent<double>(architecture, options);
// Parameterless constructor with default 10-feature, 3-action architecture
var defaultModel = new FinancialDQNAgent<double>();
Constructors
| Constructor | Summary |
|---|---|
FinancialDQNAgent() |
Initializes a new instance of the FinancialDQNAgent class. |
FinancialDQNAgent(NeuralNetworkArchitecture<T> architecture, TradingAgentOptions<T> options) |
Properties
| Property | Summary |
|---|---|
double CurrentEpsilon { get; } |
Current exploration rate: max(EpsilonEnd, EpsilonStart * EpsilonDecay^updates), where updates counts the gradient updates applied so far (the same per-update multiplicative schedule as the library’s DQNAgent). |
int FeatureCount { get; } |
Methods
| Method | Summary |
|---|---|
Dictionary<string, T> GetMetrics() |
Trading metrics plus the CURRENT exploration rate. |
Dictionary<string, T> GetTradingMetrics() |
|
ModelMetadata<T> GetModelMetadata() |
|
ModelOptions GetOptions() |
The CURRENT exploration rate, annealed from EpsilonStart toward EpsilonEnd by EpsilonDecay. |
T Train() |
|
Vector<T> ComputeGradients(Vector<T> input, Vector<T> target, ILossFunction<T> lossFunction) |
|
Vector<T> SelectAction(Vector<T> state, bool training) |
|
Vector<T> SelectAction(Vector<T> state, bool training, bool[] legalActions) |
|
byte[] Serialize() |
Auto-generated common model serialization surface. |
void ApplyGradients(Vector<T> gradients, T learningRate) |
|
void Deserialize(byte[] data) |
Auto-generated common model deserialization surface. |
void LoadModel(string filepath) |
Executes LoadModel for the FinancialDQNAgent. |
void SaveModel(string filepath) |
Executes SaveModel for the FinancialDQNAgent. |
void StoreExperience(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done) |
Executes StoreExperience for the FinancialDQNAgent. |
void StoreExperience(Vector<T> state, Vector<T> action, T reward, Vector<T> nextState, bool done, bool[] nextLegalActions) |