MyCaffe  1.12.2.41
Deep learning software for Windows C# programmers.
MyCaffe.trainers.dqn.c51.st.TrainerC51< T > Class Template Reference

The TrainerC51 implements the C51-DQN algorithm as described by Bellemare et al., Google Dopamine RainboAgent and 'flyyufelix' More...

Inheritance diagram for MyCaffe.trainers.dqn.c51.st.TrainerC51< T >:
MyCaffe.trainers.IxTrainerRL MyCaffe.trainers.IxTrainer

Public Member Functions

 TrainerC51 (MyCaffeControl< T > mycaffe, PropertySet properties, CryptoRandom random, IxTrainerCallback icallback)
 The constructor. More...
 
void Dispose ()
 Release all resources used. More...
 
bool Initialize ()
 Initialize the trainer. More...
 
bool Shutdown (int nWait)
 Shutdown the trainer. More...
 
ResultCollection RunOne (int nDelay=1000)
 Run a single cycle on the environment after the delay. More...
 
byte[] Run (int nN, PropertySet runProp, out string type)
 Run a set of iterations and return the resuts. More...
 
bool Test (int nN, ITERATOR_TYPE type)
 Run the test cycle - currently this is not implemented. More...
 
bool Train (int nN, ITERATOR_TYPE type, TRAIN_STEP step)
 Train the network using a modified PG training algorithm optimized for GPU use. More...
 

Detailed Description

The TrainerC51 implements the C51-DQN algorithm as described by Bellemare et al., Google Dopamine RainboAgent and 'flyyufelix'

See also
A Distributional Perspective on Reinforcement Learning, Marc G. Bellemare, Will Dabney, Remi Munos, 2017, arXiv:1707.06887
Dopamine: A Research Framework for Deep Reinforcement Learning Pablo Samuel Castro, Subhodeep Moitra, Carles Gelada, Saurabh Kumar, and Marc G. Bellemare, 2018, Google Brain
Github:google/dopamine, Google, 2018, license Apache 2.0 (https://github.com/google/dopamine/blob/master/LICENSE)
Github:openai/baselines, OpenAI, 2018, license MIT (https://github.com/openai/baselines/blob/master/LICENSE)
GitHub:flyyufelix/C51-DDQN-Keras 2017, MIT License (https://github.com/flyyufelix/C51-DDQN-Keras/blob/master/LICENSE)
GitHub:Kyushik/DRL, Kyushik, 2019
Github:higgsfield/RL-Adventure, higgsfield, 2018
Template Parameters
T

Definition at line 31 of file TrainerC51.cs.

Constructor & Destructor Documentation

◆ TrainerC51()

MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.TrainerC51 ( MyCaffeControl< T >  mycaffe,
PropertySet  properties,
CryptoRandom  random,
IxTrainerCallback  icallback 
)

The constructor.

Parameters
mycaffeSpecifies the MyCaffeControl to use for learning and prediction.
propertiesSpecifies the property set containing the key/value pairs of property settings.
randomSpecifies a Random number generator used for random selection.
icallbackSpecifies the callback for parent notifications and queries.

Definition at line 45 of file TrainerC51.cs.

Member Function Documentation

◆ Dispose()

Release all resources used.

Definition at line 56 of file TrainerC51.cs.

◆ Initialize()

bool MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.Initialize ( )

Initialize the trainer.

Returns
Returns true.

Implements MyCaffe.trainers.IxTrainer.

Definition at line 64 of file TrainerC51.cs.

◆ Run()

byte[] MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.Run ( int  nN,
PropertySet  runProp,
out string  type 
)

Run a set of iterations and return the resuts.

Parameters
nNSpecifies the number of samples to run.
runPropOptionally specifies properties to use when running.
typeReturns the data type contained in the byte stream.
Returns
The results of the run containing the action are returned as a byte stream.

Implements MyCaffe.trainers.IxTrainerRL.

Definition at line 122 of file TrainerC51.cs.

◆ RunOne()

ResultCollection MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.RunOne ( int  nDelay = 1000)

Run a single cycle on the environment after the delay.

Parameters
nDelaySpecifies a delay to wait before running.
Returns
The results of the run containing the action are returned.

Implements MyCaffe.trainers.IxTrainerRL.

Definition at line 106 of file TrainerC51.cs.

◆ Shutdown()

bool MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.Shutdown ( int  nWait)

Shutdown the trainer.

Parameters
nWaitSpecifies a wait in ms. for the shutdown to complete.
Returns
Returns true.

Implements MyCaffe.trainers.IxTrainer.

Definition at line 76 of file TrainerC51.cs.

◆ Test()

bool MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.Test ( int  nN,
ITERATOR_TYPE  type 
)

Run the test cycle - currently this is not implemented.

Parameters
nNSpecifies the number of iterations (based on the ITERATION_TYPE) to run, or -1 to ignore.
typeSpecifies the iteration type (default = ITERATION).
Returns
A value of true is returned when handled, false otherwise.

Implements MyCaffe.trainers.IxTrainer.

Definition at line 138 of file TrainerC51.cs.

◆ Train()

bool MyCaffe.trainers.dqn.c51.st.TrainerC51< T >.Train ( int  nN,
ITERATOR_TYPE  type,
TRAIN_STEP  step 
)

Train the network using a modified PG training algorithm optimized for GPU use.

Parameters
nNSpecifies the number of iterations (based on the ITERATION_TYPE) to run, or -1 to ignore.
typeSpecifies the iteration type (default = ITERATION).
stepSpecifies the stepping mode to use (when debugging).
Returns
A value of true is returned when handled, false otherwise.

Implements MyCaffe.trainers.IxTrainer.

Definition at line 164 of file TrainerC51.cs.


The documentation for this class was generated from the following file: