rlDQNAgent

Deep Q-network reinforcement learning agent

Description

The deep Q-network (DQN) algorithm is a model-free, online, off-policy reinforcement learning method. A DQN agent is a value-based reinforcement learning agent that trains a critic to estimate the return or future rewards. DQN is a variant of Q-learning, and it operates only within discrete action spaces.

For more information, Deep Q-Network Agents. For more information on the different types of reinforcement learning agents, see Reinforcement Learning Agents.

Creation

Syntax

agent = rlDQNAgent(observationInfo,actionInfo)

agent = rlDQNAgent(observationInfo,actionInfo,initOpts)

agent = rlDQNAgent(critic)

agent = rlDQNAgent(critic,agentOptions)

Description

Create Agent from Observation and Action Specifications

agent = rlDQNAgent(observationInfo,actionInfo) creates a DQN agent for an environment with the given observation and action specifications, using default initialization options. The critic representation in the agent uses a default multi-output Q-value deep neural network built from the observation specification observationInfo and the action specification actionInfo.

agent = rlDQNAgent(observationInfo,actionInfo,initOpts) creates a DQN agent for an environment with the given observation and action specifications. The agent uses a default network configured using options specified in the initOpts object. For more information on the initialization options, see rlAgentInitializationOptions.

Create Agent from Critic Representation

agent = rlDQNAgent(critic) creates a DQN agent with the specified critic network using a default option set for a DQN agent.

Specify Agent Options

agent = rlDQNAgent(critic,agentOptions) creates a DQN agent with the specified critic network and sets the AgentOptions property to the agentOptions input argument. Use this syntax after any of the input arguments in the previous syntaxes..

Input Arguments

observationInfo — Observation specifications

specification object | array of specification objects

Observation specifications, specified as a reinforcement learning specification object or an array of specification objects defining properties such as dimensions, data type, and names of the observation signals.

You can extract observationInfo from an existing environment or agent using getObservationInfo. You can also construct the specifications manually using rlFiniteSetSpec or rlNumericSpec.

actionInfo — Action specification

specification object

Action specifications, specified as a reinforcement learning specification object defining properties such as dimensions, data type, and names of the action signals.

Since a DDPG agent operates in a discrete action space, you must specify actionInfo as an rlFiniteSetSpec object.

You can extract actionInfo from an existing environment or agent using getActionInfo. You can also construct the specification manually using rlFiniteSetSpec.

initOpts — Agent initialization options

rlAgentInitializationOptions object

Agent initialization options, specified as an rlAgentInitializationOptions object.

critic — Critic network representation

rlQValueRepresentation object

Critic network representation, specified as an rlQValueRepresentation object. For more information on creating critic representations, see Create Policy and Value Function Representations.

Your critic representation can use a recurrent neural network as its function approximator. However, only the multi-output Q-value function representation supports recurrent neural networks. For an example, see Create DQN Agent with Recurrent Neural Network.

Properties

AgentOptions — Agent options

rlDQNAgentOptions object

Agent options, specified as an rlDQNAgentOptions object.

If you create a DQN agent with a default critic representation that uses a recurrent neural network, the default value of AgentOptions.SequenceLength is 32.

ExperienceBuffer — Experience buffer

ExperienceBuffer object

Experience buffer, specified as an ExperienceBuffer object. During training the agent stores each of its experiences (S,A,R,S') in a buffer. Here:

• S is the current observation of the environment.

• A is the action taken by the agent.

• R is the reward for taking action A.

• S' is the next observation after taking action A.

For more information on how the agent samples experience from the buffer during training, see Deep Q-Network Agents.

Object Functions

train	Train reinforcement learning agents within a specified environment
sim	Simulate trained reinforcement learning agents within specified environment
getAction	Obtain action from agent or actor representation given environment observations
getActor	Get actor representation from reinforcement learning agent
setActor	Set actor representation of reinforcement learning agent
getCritic	Get critic representation from reinforcement learning agent
setCritic	Set critic representation of reinforcement learning agent
generatePolicyFunction	Create function that evaluates trained policy of reinforcement learning agent

Examples

Create DQN Agent from Observation and Action Specifications

Create an environment with a discrete action space, and obtain its observation and action specifications. For this example, load the environment used in the example Create Agent Using Deep Network Designer and Train Using Image Observations. This environment has two observations: a 50-by-50 grayscale image and a scalar (the angular velocity of the pendulum). The action is a scalar with five possible elements (a torque of either -2, -1, 0, 1, or 2 Nm applied to a swinging pole).

% load predefined environment
env = rlPredefinedEnv("SimplePendulumWithImage-Discrete");

% obtain observation and action specifications
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

The agent creation function initializes the actor and critic networks randomly. You can ensure reproducibility by fixing the seed of the random generator. To do so, uncomment the following line.

% rng(0)

Create a deep Q-network agent from the environment observation and action specifications.

agent = rlDQNAgent(obsInfo,actInfo);

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension),rand(obsInfo(2).Dimension)})

ans = 1

You can now test and train the agent within the environment.

Create DQN Agent Using Initialization Options

Create an environment with a discrete action space, and obtain its observation and action specifications. For this example, load the environment used in the example Create Agent Using Deep Network Designer and Train Using Image Observations. This environment has two observations: a 50-by-50 grayscale image and a scalar (the angular velocity of the pendulum). The action is a scalar with five possible elements (a torque of either -2, -1, 0, 1, or 2 Nm applied to a swinging pole).

% load predefined environment
env = rlPredefinedEnv("SimplePendulumWithImage-Discrete");

% obtain observation and action specifications
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create an agent initialization option object, specifying that each hidden fully connected layer in the network must have 128 neurons (instead of the default number, 256).

initOpts = rlAgentInitializationOptions('NumHiddenUnit',128);

The agent creation function initializes the actor and critic networks randomly. You can ensure reproducibility by fixing the seed of the random generator. To do so, uncomment the following line.

% rng(0)

Create a policy gradient agent from the environment observation and action specifications.

agent = rlPGAgent(obsInfo,actInfo,initOpts);

Reduce the critic learning rate to 1e-3.

critic = getCritic(agent);
critic.Options.LearnRate = 1e-3;
agent  = setCritic(agent,critic);

Extract the deep neural network from both the critic.

criticNet = getModel(getCritic(agent));

The default DQN agent uses a multi-output Q-value critic approximator. A multi-output approximator has observations as inputs and state-action values as outputs. Each output element represents the expected cumulative long-term reward for taking the corresponding discrete action from the state indicated by the observation inputs.

Display the layers of the critic network, and verify that each hidden fully connected layer has 128 neurons

criticNet.Layers

ans = 
  11x1 Layer array with layers:

     1   'input_1'        Image Input       50x50x1 images
     2   'conv_1'         Convolution       64 3x3x1 convolutions with stride [1  1] and padding [0  0  0  0]
     3   'relu_input_1'   ReLU              ReLU
     4   'fc_1'           Fully Connected   128 fully connected layer
     5   'input_2'        Feature Input     1 features
     6   'fc_2'           Fully Connected   128 fully connected layer
     7   'concat'         Concatenation     Concatenation of 2 inputs along dimension 1
     8   'relu_body'      ReLU              ReLU
     9   'fc_body'        Fully Connected   128 fully connected layer
    10   'body_output'    ReLU              ReLU
    11   'output'         Fully Connected   1 fully connected layer

Plot the critic network

plot(layerGraph(criticNet))

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension),rand(obsInfo(2).Dimension)})

ans = 1x1 cell array
    {[2]}

You can now test and train the agent within the environment.

Create a DQN Agent Using a Multi-Output Critic Representation

Create an environment interface and obtain its observation and action specifications. For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

% load predefined environment
env = rlPredefinedEnv("CartPole-Discrete");

% get observation and action specification objects
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

For an agent with a discrete action space, you have the option to create a multi-output critic representation, which is generally more efficient than a comparable single-output critic representation.

A multi-output critic has only the observation as input, and an output vector having as many elements as the number of possible discrete actions. Each output element represents the expected cumulative long-term reward following from the observation given as input, when the corresponding discrete action is taken.

Create the multi-output critic representation using a deep neural network approximator.

% create a deep neural network approximator 
% the observation input layer must have 4 elements (obsInfo.Dimension(1))
% the action output layer must have 2 elements (length(actInfo.Elements))
dnn = [
    imageInputLayer([obsInfo.Dimension(1) 1 1], 'Normalization', 'none', 'Name', 'state')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC1')
    reluLayer('Name', 'CriticRelu1')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC2')
    reluLayer('Name','CriticCommonRelu')
    fullyConnectedLayer(length(actInfo.Elements), 'Name', 'output')];

% set some options for the critic
criticOpts = rlRepresentationOptions('LearnRate',0.01,'GradientThreshold',1);

% create the critic based on the network approximator
critic = rlQValueRepresentation(dnn,obsInfo,actInfo,'Observation',{'state'},criticOpts);

Specify agent options, and create a DQN agent using the critic.

agentOpts = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...    
    'TargetUpdateMethod',"periodic", ...
    'TargetUpdateFrequency',4, ...   
    'ExperienceBufferLength',100000, ...
    'DiscountFactor',0.99, ...
    'MiniBatchSize',256);

agent = rlDQNAgent(critic,agentOpts)

agent = 
  rlDQNAgent with properties:

        AgentOptions: [1x1 rl.option.rlDQNAgentOptions]
    ExperienceBuffer: [1x1 rl.util.ExperienceBuffer]

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(4,1)})

ans = 10

You can now test and train the agent against the environment.

Create a DQN Agent Using a Single-Output Critic Representation

Create an environment interface and obtain its observation and action specifications. For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

% load predefined environment
env = rlPredefinedEnv("CartPole-Discrete");

% get observation and specification info
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create a single-output critic representation using a deep neural network approximator. It must have both observations and action as input layers, and a single scalar output representing the expected cumulative long-term reward following from the given observation and action.

% create a deep neural network approximator 
% the observation input layer must have 4 elements (obsInfo.Dimension(1))
% the action input layer must have 1 element (actInfo.Dimension(1))
% the output must be a scalar
statePath = [
    featureInputLayer(obsInfo.Dimension(1), 'Normalization', 'none', 'Name', 'state')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC1')
    reluLayer('Name', 'CriticRelu1')
    fullyConnectedLayer(24, 'Name', 'CriticStateFC2')];
actionPath = [
    featureInputLayer(actInfo.Dimension(1), 'Normalization', 'none', 'Name', 'action')
    fullyConnectedLayer(24, 'Name', 'CriticActionFC1')];
commonPath = [
    additionLayer(2,'Name', 'add')
    reluLayer('Name','CriticCommonRelu')
    fullyConnectedLayer(1, 'Name', 'output')];
criticNetwork = layerGraph(statePath);
criticNetwork = addLayers(criticNetwork, actionPath);
criticNetwork = addLayers(criticNetwork, commonPath);    
criticNetwork = connectLayers(criticNetwork,'CriticStateFC2','add/in1');
criticNetwork = connectLayers(criticNetwork,'CriticActionFC1','add/in2');

% set some options for the critic
criticOpts = rlRepresentationOptions('LearnRate',0.01,'GradientThreshold',1);

% create the critic based on the network approximator
critic = rlQValueRepresentation(criticNetwork,obsInfo,actInfo,...
    'Observation',{'state'},'Action',{'action'},criticOpts);

Specify agent options, and create a DQN agent using the critic.

agentOpts = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...    
    'TargetUpdateMethod',"periodic", ...
    'TargetUpdateFrequency',4, ...   
    'ExperienceBufferLength',100000, ...
    'DiscountFactor',0.99, ...
    'MiniBatchSize',256);

agent = rlDQNAgent(critic,agentOpts)

agent = 
  rlDQNAgent with properties:

        AgentOptions: [1x1 rl.option.rlDQNAgentOptions]
    ExperienceBuffer: [1x1 rl.util.ExperienceBuffer]

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(4,1)})

ans = 10

You can now test and train the agent against the environment.

Create DQN Agent with Recurrent Neural Network

For this example load the predefined environment used for the Train DQN Agent to Balance Cart-Pole System example.

env = rlPredefinedEnv('CartPole-Discrete');

Get observation and action information. This environment has a continuous four-dimensional observation space (the positions and velocities of both cart and pole) and a discrete one-dimensional action space consisting on the application of two possible forces, -10N or 10N.

obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create a recurrent deep neural network for your critic. To create a recurrent neural network, use a sequenceInputLayer as the input layer and include an lstmLayer as one of the other network layers.

For DQN agents, only the multi-output Q-value function representation supports recurrent neural networks. For multi-output Q-value function representations, the number of elements of the output layer has to be equal to the number of possible actions: numel(actInfo.Elements).

criticNetwork = [
    sequenceInputLayer(obsInfo.Dimension(1),'Normalization','none','Name','state')
    fullyConnectedLayer(50, 'Name', 'CriticStateFC1')
    reluLayer('Name','CriticRelu1')
    lstmLayer(20,'OutputMode','sequence','Name','CriticLSTM');
    fullyConnectedLayer(20,'Name','CriticStateFC2')
    reluLayer('Name','CriticRelu2')
    fullyConnectedLayer(numel(actInfo.Elements),'Name','output')];

Create a representation for your critic using the recurrent neural network.

criticOptions = rlRepresentationOptions('LearnRate',1e-3,'GradientThreshold',1);
critic = rlQValueRepresentation(criticNetwork,obsInfo,actInfo,...
    'Observation','state',criticOptions);

Specify options for creating the DQN agent. To use a recurrent neural network, you must specify a SequenceLength greater than 1.

agentOptions = rlDQNAgentOptions(...
    'UseDoubleDQN',false, ...
    'TargetSmoothFactor',5e-3, ...
    'ExperienceBufferLength',1e6, ...
    'SequenceLength',20);
agentOptions.EpsilonGreedyExploration.EpsilonDecay = 1e-4;

Create the agent. The actor and critic networks are initialized randomly.

agent = rlDQNAgent(critic,agentOptions);

Check your agent using getAction to return the action from a random observation.

getAction(agent,rand(obsInfo.Dimension))

ans = -10