Classes
struct	A2CTrainerCInfo
	Data used to construct an IA2CTrainer instance More...

struct	ActivationNodeCInfo
	Data used to construct an activation node instance More...

struct	AgentDataStoreCInfo
	Data used to construct an IAgentDataStore instance More...

struct	BufferInputCInfo
	Data used to construct an IBufferInput instance More...

struct	ChoiceNodeCInfo
	Data used to construct a choice node instance More...

struct	ComponentInputBinding
	Dynamically binds a node to a component input. More...

struct	ComponentInputCInfo
	Data used to construct a component input instance More...

struct	ConcatNodeCInfo
	Data used to construct a concat node instance More...

struct	Conv2DInfo

struct	Conv2DLayerCInfo
	Conv2D layer constructor info More...

struct	CuriosityModuleCInfo
	Data used to construct an ICuriosityModule instance More...

struct	D4PGTrainerCInfo
	Data used to construct an ID4PGTrainer instance More...

struct	ExtractNodeCInfo
	Data used to construct an extract node instance More...

struct	GeneSwapInfo
	GeneticTrainer gene swap info More...

struct	GeneticAgentTrainerCInfo
	Data used to construct an IGeneticAgentTrainer instance More...

struct	GeneticTrainerCInfo
	Data used to construct an IA2CTrainer instance More...

struct	GeneticTrainerInitializationInfo
	Initializes the GeneticTrainer with a new population More...

struct	GeneticTrainingInfo
	GeneticTrainer parameter info More...

struct	GradientDescentTrainerCInfo
	Data used to construct an IGradientDescentTrainer instance More...

struct	GradientDescentTrainingInfo
	GradientDescentTrainer training info More...

struct	GraphCInfo
	Data used to construct an IGraph instance More...

struct	GraphComponentCInfo
	Data used to construct an IGraphComponent instance More...

struct	GraphControllerCInfo
	Data used to construct an IGraphController instance More...

struct	GraphInputCInfo
	Data used to construct an IGraphInput instance More...

struct	GraphManagerCInfo
	Data used to construct an IGraphManager instance More...

struct	GraphNodeCInfo
	Data used to construct an IGraphNode instance More...

class	IA2CTrainer
	The A2C Trainer is a reinforcement learning trainer that is composed of two parts: an actor sub graph and a critic sub-graph. More...

class	IAgent
	Agents are used to track the performance of one instance of a network. More...

class	IAgentDataStore
	The agent data store keeps experience data for the purpose of training. Some RL trainers don't store long term data in the store, while others keep a long history of data. More...

class	IAgentFactory
	RL trainers implement this to create agents. More...

class	IBufferInput
	Input node into the AI graph. Allows for specifying a tensor input. More...

class	IComponentInput
	Input node into a component. At the time of component construction, a node can be fed into this input. The incoming node must have the same dimension as this node. This node will then act as a passthrough for the incoming node. In this way, the component can be used at any point in the graph; not just the beginning. More...

class	IContext
	Every node in the AI graph must belong to the same context. More...

class	IConv2DLayer
	2D convolution layer. Input is interpreted as a 2D grid laid in row major order. Filters are applied to blocks of the input grid at a time, producing another 2D grid as output. This 2D grid can be fed into neuron layers where it will be treated as a 1D array. More...

class	ICuriosityModule
	A curiosity module is a way of rewarding an agent for behavior not yet seen. Rewards are given based on how expected the next set of observations are given the previous observations and the chosen actions. Unexpected results are given larger rewards than behavior the agent saw in the past, thus promoting the agent to be curious and explore the world. More...

class	ID4PGTrainer
	The D4PGTrainer is a reinforcement learning trainer that is composed of two parts: an actor sub graph and a critic sub-graph. Unlike A2C and PPO, the critic graph is created and managed internally by SmartEngine. You only need to supply the actor graph. More...

class	IGeneticAgentTrainer
	A genetic trainer that uses agents to automatically set the loss of each chromosome. Agents should be mapped to a chromosome after creation by calling MapAgentToChromosome() More...

class	IGeneticTrainer
	Trains a NeuronLayer / NeuralNetwork using a genetic algorithm. With each step, networks are sorted by loss. The top networks are left unchanged, while the bottom networks are replaced by new ones. The new networks are formed by mixing the weights of the top networks and randomly mutating weights / neurons. Lastly, the bottom percent of the networks can be replaced with completely random weights, adding some variability to the pool. More...

class	IGradientDescentTrainer
	Trains a set of networks using gradient descent. Training with GradientDescentTrainer requires a Loss structure and thus the output of the network must be known. If only a relative loss about the network is known, a GeneticTrainer needs to be used. More...

class	IGraph
	A graph is a collection of buffers and nodes that together form a neural network. The graph is created from a json definition file. After creation, the contents of the graph can be loaded from or saved to disk. More...

class	IGraphComponent
	A graph component allows graphs to embed other graphs. Nodes within components can be referenced using the following syntax: [Component Name]:[Component Graph Node Name] References can be chained together to dive to sub components. More...

class	IGraphController
	Represents a use of a graph. Owners of the controller feed data to the controller at regular intervals. When the graph model is executed, the controller will have data that can be acted upon More...

class	IGraphInput
	Base class for input nodes into the AI graph More...

class	IGraphInputOutput
	Views on top of models deal with this class to inject and extract data from the underlying model. More...

class	IGraphManager
	Connects graph controllers with graph models. Graph controllers define a single use of a graph. They pipe data to the model and later act on the output of the model. The model (usually a graph, but can be a dummy class) has the job of aggregating the input from one or more controllers and producing an output for each. More...

class	IGraphModel
	Takes input from a controller and produces an output. If using a graph as the model, this interface does not need to be implemented. However, you can provide an instance of this interface to create "dummy" models. More...

class	IGraphModelOutput
	Used to return output back to controllers. More...

class	IGraphNode
	A logical node in the AI graph. Some nodes, like NeuralNetwork, are composed of other nodes (neuron layers). More...

class	ILoss
	The loss of a NeuralNetwork is computed using the formula (Expected Ouput - Actual Output)^2 The mean of all these values across the output tensor is used when crunching down the loss to a single value. More...

class	ILossTrainer
	Base class for NeuralNetwork loss trainers More...

class	IMatrix
	Wrapper around either a const or a mutable matrix. More...

class	IMemoryBuffer
	Simple in memory buffer. Used for serialization, but can be used as a generic buffer too. More...

class	INeuronLayer
	A neuron layer is the trainable unit in the neural network graph. These can chained together to allow the network to learn and represent complex relationships. More...

class	IObject
	Base class for SmartEngine AI objects. It is not common to deal with this class directly. More...

class	IParameter
	A parameter is similar to a NeuronLayer in that it is trainable. However, it takes no input and is simply a layer of weights that can be used as input into other parts of the graph. More...

class	IPoolingLayer
	2D pooling layer. Typically used after a conv2d layer to reduce the output dimension. Fixed size of 2x2 blocks. More...

class	IPPOTrainer
	The PPO Trainer is a reinforcement learning trainer that is composed of two parts: an actor sub graph and a critic sub-graph. One of the differences between PPO and A2C is that PPO works off of large batches of data. Instead of waiting for a small batch before training, PPO will wait until there is a mega batch, called the trajectory. When we've received that much data, we train over it in normal batch sizes for a couple of times and then wait for the next trajectory. More...

class	IResource
	Base class for objects that can be loaded from and saved to disk. More...

class	IRLTrainer
	Base class for all reinforcement learning trainers. More...

class	ISerializable
	Base class for objects that can be loaded from and saved to an in memory buffer. More...

class	IStandardGraphModel
	Interface to the standard model that interacts with a graph. This can be used to extend the standard model with custom code by encapsulating this object More...

class	IUnpoolingLayer
	2D unpooling layer. Typically used before a conv2d transpose layer to expand the output dimension. Fixed size of 2x2 blocks. More...

struct	LayerInfo
	Layer information More...

struct	LossCInfo
	Data used to construct an ILoss instance More...

struct	LossTrainerCInfo
	Data used to construct an ILossTrainer instance. More...

struct	LossTrainingMethodInfo

struct	MaximumNodeCInfo
	Data used to construct a maximum node instance More...

struct	MinimumNodeCInfo
	Data used to construct a minimum node instance More...

struct	MultiplexerNodeCInfo
	Data used to construct a multiplexer node instance. More...

struct	MutationInfo
	GeneticTrainer mutation info More...

struct	NeuronLayerCInfo
	Data used to construct an INeuronLayer instance More...

struct	NormalNodeCInfo
	Data used to construct a normal node instance More...

class	ObjectPtr
	Smart pointer to an IObject. Automatic ref counting. More...

struct	ParameterCInfo
	Data used to construct an IParameter instance More...

struct	PoolingLayerCInfo
	Pool constructor info More...

struct	PPOTrainerCInfo
	Data used to construct an IPPOTrainer instance More...

struct	RegularizationLossInfo
	Regularization tries to keep the weight values from exploding during gradient descent by adding L1 and L2 loss of the weights. More...

struct	ResourceCInfo
	Data used to construct an IResource instance More...

struct	RLTrainerCInfo
	Data used to construct an IRLTrainer instance More...

struct	StandardGraphModelCInfo
	Data used to construct an IStandardGraphModel instance More...

struct	StopGradientNodeCInfo
	Data used to construct a stop gradient node instance More...

class	StrongNumber

struct	UnpoolingLayerCInfo
	Unpool constructor info More...

struct	UserAgentCInfo
	Data used to construct an overridable instance of IAgent More...

struct	UserGraphModelCInfo
	Data used to construct an overridable instance of IGraphModel More...

Typedefs
typedef void(SMARTENGINE_CALLBACK *	Agent_StartNewExperience_Ptr) (void *userData)

typedef void(SMARTENGINE_CALLBACK *	Agent_RecordObservation_Ptr) (void userData, const char inputName, const float *buffer, int64 bufferCount)

typedef void(SMARTENGINE_CALLBACK *	Agent_RecordAction_Ptr) (void userData, const char actionName, const float *buffer, int64 bufferCount)

typedef void(SMARTENGINE_CALLBACK *	Agent_GiveReward_Ptr) (void *userData, float reward)

typedef void(SMARTENGINE_CALLBACK *	Agent_EndEpisode_Ptr) (void *userData)

typedef bool(*	AssetStoreHandler) (const char assetName, AssetType assetType, IMemoryBuffer buffer)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_ClearData_Ptr) (void *userData)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_AddData_Ptr) (void userData, const char nodeName, ObjPtr data)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_RequestOutput_Ptr) (void userData, int controllerId, const char nodeName, int64 expectedRowCount)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_Execute_Ptr) (void *userData)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_RetrieveOutput_Ptr) (void *userData, int controllerId, ObjPtr modelOut)

typedef void(SMARTENGINE_CALLBACK *	GraphModel_Reset_Ptr) (void *userData)

typedef unsigned int	HashKey

using	MatrixIndex = StrongNumber< struct MatrixIndexParam, std::ptrdiff_t >

typedef int64	ObjectId

typedef unsigned int	ObjectClassId

Enumerations
enum	ActivationType : byte { ActivationType::None, ActivationType::Sigmoid, ActivationType::Tanh, ActivationType::Relu, ActivationType::Relu6, ActivationType::LeakyRelu, ActivationType::Softmax, ActivationType::Elu, ActivationType::Selu, ActivationType::Softplus, ActivationType::Normalize, ActivationType::NormalizeAllowSmallLengths, ActivationType::Count }
	The activation function to apply to the output of the NeuronLayer More...

enum	AssetType : byte { AssetType::JsonString, AssetType::Binary, AssetType::Count }

enum	ChoiceType : byte { ChoiceType::Random, ChoiceType::Max, ChoiceType::Count }
	Defines the descision algorithm of the ChoiceNode More...

enum	DeviceType : byte { DeviceType::Default, DeviceType::Cpu, DeviceType::Gpu }
	The type of context to create. More...

enum	Conv2DType : byte { Conv2DType::Conv2D, Conv2DType::Conv2DWithBias, Conv2DType::Conv2DTranspose, Conv2DType::Conv2DTransposeWithBias, Conv2DType::Count }

enum	GeneticAgentScoringMethod : byte { GeneticAgentScoringMethod::Sum, GeneticAgentScoringMethod::Average }
	Defines how the trainer deals with chromosome scores from multiple agents. More...

enum	MutationTarget : byte { MutationTarget::Auto, MutationTarget::Weight, MutationTarget::Neuron }

enum	GeneSwapTarget : byte { GeneSwapTarget::Weight, GeneSwapTarget::Neuron }

enum	GradientDescentTrainingAlgorithm : byte { GradientDescentTrainingAlgorithm::Adam, GradientDescentTrainingAlgorithm::AdaMax }

enum	LogLevel { LogLevel::Info, LogLevel::Warning, LogLevel::Error, LogLevel::Count }

enum	LossTrainingMethod : byte { LossTrainingMethod::WholeDataset, LossTrainingMethod::Stochastic }

enum	LayerType : byte { LayerType::LinearNoBias, LayerType::Linear, LayerType::LSTM, LayerType::Count }
	The type of layer to create More...

enum	PoolType : byte { PoolType::Max, PoolType::Average, PoolType::Count }

enum	SerializationResult : byte { SerializationResult::Success, SerializationResult::UnspecifiedError, SerializationResult::EndOfStream, SerializationResult::NotSupported, SerializationResult::Corrupted, SerializationResult::VersionTooLow, SerializationResult::InvalidResourceName, SerializationResult::InvalidObject, SerializationResult::InvalidBuffer, SerializationResult::InvalidInput, SerializationResult::InputCountMismatch, SerializationResult::InputDimensionMismatch, SerializationResult::TensorCountMismatch, SerializationResult::TensorSizeMismatch, SerializationResult::TensorDataSizeMismatch, SerializationResult::NodeCountMismatch, SerializationResult::NodeNotFound, SerializationResult::FileNotFound, SerializationResult::SignatureNotValid, SerializationResult::LicenseExpired, SerializationResult::InvalidGraphDefinition }

Functions
SMARTENGINE_EXPORT ObjectPtr< IA2CTrainer >	CreateA2CTrainer (const A2CTrainerCInfo &cinfo)
	Creates an instance of IA2CTrainer More...

SMARTENGINE_EXPORT ObjectPtr< IAgentDataStore >	CreateAgentDataStore (const AgentDataStoreCInfo &cinfo)
	Creates an instance of IAgentDataStore More...

SMARTENGINE_EXPORT ObjectPtr< IAgent >	CreateUserAgent (const UserAgentCInfo &cinfo)
	Creates an overridable instance of IAgent from a previously serialized agent. More...

SMARTENGINE_EXPORT void	SetAssetStoreHandlers (AssetStoreHandler loadHandler, AssetStoreHandler saveHandler)
	Used to set the handler for custom loading and saving of assets. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateConcatNode (const ConcatNodeCInfo &cinfo)
	Merges two or more node streams into one. Each input stream must have the same number of rows. The output dimension is the sum of the input dimensions. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateExtractNode (const ExtractNodeCInfo &cinfo)
	Extracts a subset of columns from the input node. The start index + column count must not exceed the output dimension of the input node. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateChoiceNode (const ChoiceNodeCInfo &cinfo)
	Returns a random choice (as an integer) from the input node's probability distribution on each call to RetrieveOutput(). More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateActivationNode (const ActivationNodeCInfo &cinfo)
	Applies an activation function to the input. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateNormalNode (const NormalNodeCInfo &cinfo)
	Returns a random value from the normal distribution given by the input nodes. NOTE: Input takes in the variance and not the standard deviation. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateMinimumNode (const MinimumNodeCInfo &cinfo)
	Takes the minimum of the inputs. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateMaximumNode (const MaximumNodeCInfo &cinfo)
	Takes the maximum of the inputs. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateStopGradientNode (const StopGradientNodeCInfo &cinfo)
	Stops the gradient from flowing backwards from this point. Only meaningful when training with gradient descent or reinforcement learning. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphNode >	CreateMultiplexerNode (const MultiplexerNodeCInfo &cinfo)
	Uses the value from the selector to decide which of the inputs to pass through. The selected input will be transmitted without adjustment. Each input should have the same dimension. The selector should have a dimension of 1 and a sigmoid activation. More...

SMARTENGINE_EXPORT ObjectPtr< IContext >	CreateContext (DeviceType device=DeviceType::Default)
	Creates a context using the specified device. More...

SMARTENGINE_EXPORT ObjectPtr< IConv2DLayer >	CreateConv2DLayer (const Conv2DLayerCInfo &cinfo)
	Creates an instance of IConv2DLayer More...

SMARTENGINE_EXPORT ObjectPtr< ICuriosityModule >	CreateCuriosityModule (const CuriosityModuleCInfo &cinfo)
	Creates an instance of IA2CTrainer More...

SMARTENGINE_EXPORT ObjectPtr< ID4PGTrainer >	CreateD4PGTrainer (const D4PGTrainerCInfo &cinfo)
	Creates an instance of ID4PGTrainer More...

SMARTENGINE_EXPORT ObjectPtr< IGeneticAgentTrainer >	CreateGeneticAgentTrainer (const GeneticAgentTrainerCInfo &cinfo)
	Creates an instance of IGeneticAgentTrainer More...

SMARTENGINE_EXPORT ObjectPtr< IGeneticTrainer >	CreateGeneticTrainer (const GeneticTrainerCInfo &cinfo)
	Creates an instance of IGeneticTrainer More...

SMARTENGINE_EXPORT ObjectPtr< IGradientDescentTrainer >	CreateGradientDescentTrainer (const GradientDescentTrainerCInfo &cinfo)
	Creates an instance of IGradientDescentTrainer More...

SMARTENGINE_EXPORT ObjectPtr< IGraph >	CreateEmptyGraph (const GraphCInfo &cinfo)
	Creates a new, empty IGraph instance. More...

SMARTENGINE_EXPORT ObjectPtr< IGraph >	CreateGraphFromDefinition (const GraphCInfo &cinfo, const char *graphDefinition)
	Creates a new IGraph from a graph definition json string. More...

SMARTENGINE_EXPORT ObjectPtr< IGraph >	CreateGraphFromResource (const GraphCInfo &cinfo, const char *resourceName=nullptr)
	Creates and loads a new IGraph from a resource. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphComponent >	CreateGraphComponent (const GraphComponentCInfo &cinfo)
	Creates an instance of IGraphComponent More...

SMARTENGINE_EXPORT ObjectPtr< IBufferInput >	CreateBufferInput (const BufferInputCInfo &cinfo)
	Creates an instance of IBufferInput More...

SMARTENGINE_EXPORT ObjectPtr< IComponentInput >	CreateComponentInput (const ComponentInputCInfo &cinfo)
	Creates an instance of IComponentInput More...

SMARTENGINE_EXPORT ObjectPtr< IGraphController >	CreateGraphController (const GraphControllerCInfo &cinfo)
	Creates an instance of IGraphController. More...

SMARTENGINE_EXPORT ObjectPtr< IStandardGraphModel >	CreateStandardGraphModel (const StandardGraphModelCInfo &cinfo)
	Creates an instance of IStandardGraphModel. More...

SMARTENGINE_EXPORT ObjectPtr< IGraphModel >	CreateUserGraphModel (const UserGraphModelCInfo &cinfo)
	Creates an overridable instance if IGraphModel More...

SMARTENGINE_EXPORT ObjectPtr< IGraphManager >	CreateGraphManager (const GraphManagerCInfo &cinfo)
	Creates an instance of IGraphManager. More...

SMARTENGINE_EXPORT HashKey	HashString (const char *str)

SMARTENGINE_EXPORT bool	SetSmartEngineLicense (const byte *data, int64 count)
	Loads a license from a license file blob. Call this before creating a context if you want to load a license from other than the default location. More...

SMARTENGINE_EXPORT void	SetLogHandler (LogHandler handler)
	Used to set the handler for log output from the SmartEngine AI plugin More...

SMARTENGINE_EXPORT ObjectPtr< ILoss >	CreateLoss (const LossCInfo &cinfo)
	Creates an instance of ILoss. More...

SMARTENGINE_EXPORT ObjectPtr< IMatrix >	CreateMatrix (MatrixIndex rows, MatrixIndex cols)
	Creates an instance of IMatrix More...

SMARTENGINE_EXPORT ObjectPtr< IMemoryBuffer >	CreateMemoryBuffer ()
	Creates an instance of IMemoryBuffer More...

SMARTENGINE_EXPORT ObjectPtr< INeuronLayer >	CreateNeuronLayer (const NeuronLayerCInfo &cinfo)
	Creates an instance of INeuronLayer More...

template<typename D >
ObjectPtr< D >	DynamicCast (IObject *p)
	Safe cast an IObject to a pointer of a different type. More...

template<typename D >
ObjectPtr< D >	DynamicCast (const IObject *p)
	Safe cast an IObject to a pointer of a different type. More...

template<typename D >
ObjectPtr< D >	DynamicCastAttach (IObject *p)
	Safe cast an IObject to a pointer of a different type. Does not call AddRef() on the interface. More...

template<typename D >
ObjectPtr< D >	IsA (IObject *p)

SMARTENGINE_EXPORT ObjectPtr< IParameter >	CreateParameter (const ParameterCInfo &cinfo)
	Creates an instance of IParameter More...

SMARTENGINE_EXPORT ObjectPtr< IPoolingLayer >	CreatePoolingLayer (const PoolingLayerCInfo &cinfo)
	Creates an instance of IPoolingLayer More...

SMARTENGINE_EXPORT ObjectPtr< IUnpoolingLayer >	CreateUnpoolingLayer (const UnpoolingLayerCInfo &cinfo)
	Creates an instance of IUnpoolingLayer More...

SMARTENGINE_EXPORT ObjectPtr< IPPOTrainer >	CreatePPOTrainer (const PPOTrainerCInfo &cinfo)
	Creates an instance of IPPOTrainer More...

Variables
const float	cUpdateEveryFrameRate = 0.0f
	An update rate set to this value means we should update the controller with model output every frame. More...

const float	cUpdateManuallyRate = -1.0f
	An update rate set to this value means we shouldn't update the controller automatically, but rather only when UpdateControllerImmediately() is called. More...

const float	cDefaultUpdateRate = -2.0f
	Use the model's default update rate. Not valid when registering a model. This is the only value that can be used to group controllers to the same network. More...

Typedef Documentation

◆ Agent_EndEpisode_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::Agent_EndEpisode_Ptr) (void *userData)

◆ Agent_GiveReward_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::Agent_GiveReward_Ptr) (void *userData, float reward)

◆ Agent_RecordAction_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::Agent_RecordAction_Ptr) (void *userData, const char *actionName, const float *buffer, int64 bufferCount)

◆ Agent_RecordObservation_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::Agent_RecordObservation_Ptr) (void *userData, const char *inputName, const float *buffer, int64 bufferCount)

◆ Agent_StartNewExperience_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::Agent_StartNewExperience_Ptr) (void *userData)

◆ AssetStoreHandler

typedef bool(* SmartEngine::AssetStoreHandler) (const char *assetName, AssetType assetType, IMemoryBuffer *buffer)

◆ GraphModel_AddData_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_AddData_Ptr) (void *userData, const char *nodeName, ObjPtr data)

◆ GraphModel_ClearData_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_ClearData_Ptr) (void *userData)

◆ GraphModel_Execute_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_Execute_Ptr) (void *userData)

◆ GraphModel_RequestOutput_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_RequestOutput_Ptr) (void *userData, int controllerId, const char *nodeName, int64 expectedRowCount)

◆ GraphModel_Reset_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_Reset_Ptr) (void *userData)

◆ GraphModel_RetrieveOutput_Ptr

typedef void(SMARTENGINE_CALLBACK * SmartEngine::GraphModel_RetrieveOutput_Ptr) (void *userData, int controllerId, ObjPtr modelOut)

◆ HashKey

typedef unsigned int SmartEngine::HashKey

◆ MatrixIndex

using SmartEngine::MatrixIndex = typedef StrongNumber<struct MatrixIndexParam, std::ptrdiff_t>

◆ ObjectClassId

typedef unsigned int SmartEngine::ObjectClassId

◆ ObjectId

typedef int64 SmartEngine::ObjectId

Enumeration Type Documentation

◆ ActivationType

enum SmartEngine::ActivationType : byte

strong

The activation function to apply to the output of the NeuronLayer

Enumerator
None	No activation
Sigmoid	Maps output into the range [0..1]
Tanh	Maps output into the range [-1..1]
Relu	max(x, 0)
Relu6	min(max(x, 0), 6)
LeakyRelu	Similar to Relu, but allows for scaled negative values
Softmax	Converts the outputs into weighted probabilities.
Elu	Similar to Relu, but uses an exponential for negative values
Selu	Similar to Relu, but uses an exponential for negative values. Positive values are scaled. Best when training with gradient descent. The inputs to the network should be be in the [-2..2] range.
Softplus	Maps output into a positive number
Normalize	Treats each row as a vector that is to have total length 1.0 If the length of the row is 0.0, the row is unchanged
NormalizeAllowSmallLengths	Treats each row as a vector. Vector lengths greater than 1.0 will be normalized to length 1.0. Vector lengths less than 1.0 will be passed unchanged.
Count

◆ AssetType

enum SmartEngine::AssetType : byte

strong

Enumerator
JsonString	The memory buffer will be filled or should be filled with just a single string that is the content of the asset.
Binary	The memory buffer will be filled or should be filled with the raw bytes from the content of the asset.
Count

◆ ChoiceType

enum SmartEngine::ChoiceType : byte

strong

Defines the descision algorithm of the ChoiceNode

Enumerator
Random	Treat the output as a probability distribution and sample from that distribution on each evaulation. Returns the index of the chosen output.
Max	Return the index of the output with the highest value.
Count

◆ Conv2DType

enum SmartEngine::Conv2DType : byte

strong

Enumerator
Conv2D	2D Convolution layer, no bias
Conv2DWithBias	2D Convolution layer, with bias
Conv2DTranspose	2D Transpose convolution layer (sometimes called deconvolution), no bias
Conv2DTransposeWithBias	2D Transpose convolution layer (sometimes called deconvolution), with bias
Count

◆ DeviceType

enum SmartEngine::DeviceType : byte

strong

The type of context to create.

Enumerator
Default	Chooses the best device for the current system.
Cpu	Train and evaulate using only the CPU.
Gpu	Train and evaulate using only the GPU.

◆ GeneSwapTarget

enum SmartEngine::GeneSwapTarget : byte

strong

Enumerator
Weight	Swapping occurs at the individual weight level across all neurons.
Neuron	Swapping applies to all weights in a neuron.

◆ GeneticAgentScoringMethod

enum SmartEngine::GeneticAgentScoringMethod : byte

strong

Defines how the trainer deals with chromosome scores from multiple agents.

Enumerator
Sum	Sum the scores of all agents associated with the chromosome
Average	Average the scores of all agents associated with the chromosome

◆ GradientDescentTrainingAlgorithm

enum SmartEngine::GradientDescentTrainingAlgorithm : byte

strong

Enumerator
Adam	Good default choice
AdaMax	Can be more stable in certain circumstances, such as RL training

◆ LayerType

enum SmartEngine::LayerType : byte

strong

The type of layer to create

Enumerator
LinearNoBias	Linear with no bias (X * W)
Linear	Linear with bias (X * W + B)
LSTM	Long short term memory layer
Count

◆ LogLevel

enum SmartEngine::LogLevel

strong

Enumerator
Info	Diagnostic or informational messages
Warning	Warning - recoverable error
Error	Unrecoverable error
Count

◆ LossTrainingMethod

enum SmartEngine::LossTrainingMethod : byte

strong

Enumerator
WholeDataset	Train against all data in the buffer in groups of size BatchSize.
Stochastic	Train by random sampling of the input data in groups of size BatchSize.

◆ MutationTarget

enum SmartEngine::MutationTarget : byte

strong

Enumerator
Auto	Chooses the default value depending on the node type.
Weight	Mutation occurs at the individual weight level across all neurons.
Neuron	Mutation applies to all weights in a neuron.

◆ PoolType

enum SmartEngine::PoolType : byte

strong

Enumerator
Max	Outputs the maximum value of the block of inputs
Average	Takes the average value of the block of inputs
Count

◆ SerializationResult

enum SmartEngine::SerializationResult : byte

strong

Enumerator
Success
UnspecifiedError
EndOfStream
NotSupported
Corrupted
VersionTooLow
InvalidResourceName
InvalidObject
InvalidBuffer
InvalidInput
InputCountMismatch
InputDimensionMismatch
TensorCountMismatch
TensorSizeMismatch
TensorDataSizeMismatch
NodeCountMismatch
NodeNotFound
FileNotFound
SignatureNotValid
LicenseExpired
InvalidGraphDefinition

Function Documentation

◆ CreateA2CTrainer()

SMARTENGINE_EXPORT ObjectPtr<IA2CTrainer> SmartEngine::CreateA2CTrainer ( const A2CTrainerCInfo & cinfo )

Creates an instance of IA2CTrainer

◆ CreateActivationNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateActivationNode ( const ActivationNodeCInfo & cinfo )

Applies an activation function to the input.

◆ CreateAgentDataStore()

SMARTENGINE_EXPORT ObjectPtr<IAgentDataStore> SmartEngine::CreateAgentDataStore ( const AgentDataStoreCInfo & cinfo )

Creates an instance of IAgentDataStore

◆ CreateBufferInput()

SMARTENGINE_EXPORT ObjectPtr<IBufferInput> SmartEngine::CreateBufferInput ( const BufferInputCInfo & cinfo )

Creates an instance of IBufferInput

◆ CreateChoiceNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateChoiceNode ( const ChoiceNodeCInfo & cinfo )

Returns a random choice (as an integer) from the input node's probability distribution on each call to RetrieveOutput().

◆ CreateComponentInput()

SMARTENGINE_EXPORT ObjectPtr<IComponentInput> SmartEngine::CreateComponentInput ( const ComponentInputCInfo & cinfo )

Creates an instance of IComponentInput

◆ CreateConcatNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateConcatNode ( const ConcatNodeCInfo & cinfo )

Merges two or more node streams into one. Each input stream must have the same number of rows. The output dimension is the sum of the input dimensions.

◆ CreateContext()

SMARTENGINE_EXPORT ObjectPtr<IContext> SmartEngine::CreateContext ( DeviceType device = DeviceType::Default )

Creates a context using the specified device.

◆ CreateConv2DLayer()

SMARTENGINE_EXPORT ObjectPtr<IConv2DLayer> SmartEngine::CreateConv2DLayer ( const Conv2DLayerCInfo & cinfo )

Creates an instance of IConv2DLayer

◆ CreateCuriosityModule()

SMARTENGINE_EXPORT ObjectPtr<ICuriosityModule> SmartEngine::CreateCuriosityModule ( const CuriosityModuleCInfo & cinfo )

Creates an instance of IA2CTrainer

◆ CreateD4PGTrainer()

SMARTENGINE_EXPORT ObjectPtr<ID4PGTrainer> SmartEngine::CreateD4PGTrainer ( const D4PGTrainerCInfo & cinfo )

Creates an instance of ID4PGTrainer

◆ CreateEmptyGraph()

SMARTENGINE_EXPORT ObjectPtr<IGraph> SmartEngine::CreateEmptyGraph ( const GraphCInfo & cinfo )

Creates a new, empty IGraph instance.

◆ CreateExtractNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateExtractNode ( const ExtractNodeCInfo & cinfo )

Extracts a subset of columns from the input node. The start index + column count must not exceed the output dimension of the input node.

◆ CreateGeneticAgentTrainer()

SMARTENGINE_EXPORT ObjectPtr<IGeneticAgentTrainer> SmartEngine::CreateGeneticAgentTrainer ( const GeneticAgentTrainerCInfo & cinfo )

Creates an instance of IGeneticAgentTrainer

◆ CreateGeneticTrainer()

SMARTENGINE_EXPORT ObjectPtr<IGeneticTrainer> SmartEngine::CreateGeneticTrainer ( const GeneticTrainerCInfo & cinfo )

Creates an instance of IGeneticTrainer

◆ CreateGradientDescentTrainer()

SMARTENGINE_EXPORT ObjectPtr<IGradientDescentTrainer> SmartEngine::CreateGradientDescentTrainer ( const GradientDescentTrainerCInfo & cinfo )

Creates an instance of IGradientDescentTrainer

◆ CreateGraphComponent()

SMARTENGINE_EXPORT ObjectPtr<IGraphComponent> SmartEngine::CreateGraphComponent ( const GraphComponentCInfo & cinfo )

Creates an instance of IGraphComponent

◆ CreateGraphController()

SMARTENGINE_EXPORT ObjectPtr<IGraphController> SmartEngine::CreateGraphController ( const GraphControllerCInfo & cinfo )

Creates an instance of IGraphController.

◆ CreateGraphFromDefinition()

SMARTENGINE_EXPORT ObjectPtr<IGraph> SmartEngine::CreateGraphFromDefinition	(	const GraphCInfo &	cinfo,
		const char *	graphDefinition
	)

Creates a new IGraph from a graph definition json string.

◆ CreateGraphFromResource()

SMARTENGINE_EXPORT ObjectPtr<IGraph> SmartEngine::CreateGraphFromResource	(	const GraphCInfo &	cinfo,
		const char *	resourceName = `nullptr`
	)

Creates and loads a new IGraph from a resource.

◆ CreateGraphManager()

SMARTENGINE_EXPORT ObjectPtr<IGraphManager> SmartEngine::CreateGraphManager ( const GraphManagerCInfo & cinfo )

Creates an instance of IGraphManager.

◆ CreateLoss()

SMARTENGINE_EXPORT ObjectPtr<ILoss> SmartEngine::CreateLoss ( const LossCInfo & cinfo )

Creates an instance of ILoss.

◆ CreateMatrix()

SMARTENGINE_EXPORT ObjectPtr<IMatrix> SmartEngine::CreateMatrix	(	MatrixIndex	rows,
		MatrixIndex	cols
	)

Creates an instance of IMatrix

◆ CreateMaximumNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateMaximumNode ( const MaximumNodeCInfo & cinfo )

Takes the maximum of the inputs.

◆ CreateMemoryBuffer()

SMARTENGINE_EXPORT ObjectPtr<IMemoryBuffer> SmartEngine::CreateMemoryBuffer ( )

Creates an instance of IMemoryBuffer

◆ CreateMinimumNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateMinimumNode ( const MinimumNodeCInfo & cinfo )

Takes the minimum of the inputs.

◆ CreateMultiplexerNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateMultiplexerNode ( const MultiplexerNodeCInfo & cinfo )

Uses the value from the selector to decide which of the inputs to pass through. The selected input will be transmitted without adjustment. Each input should have the same dimension. The selector should have a dimension of 1 and a sigmoid activation.

Note: The selector input nor its subgraph can be trained through gradient descent or reinforcement learning.

◆ CreateNeuronLayer()

SMARTENGINE_EXPORT ObjectPtr<INeuronLayer> SmartEngine::CreateNeuronLayer ( const NeuronLayerCInfo & cinfo )

Creates an instance of INeuronLayer

◆ CreateNormalNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateNormalNode ( const NormalNodeCInfo & cinfo )

Returns a random value from the normal distribution given by the input nodes. NOTE: Input takes in the variance and not the standard deviation.

◆ CreateParameter()

SMARTENGINE_EXPORT ObjectPtr<IParameter> SmartEngine::CreateParameter ( const ParameterCInfo & cinfo )

Creates an instance of IParameter

◆ CreatePoolingLayer()

SMARTENGINE_EXPORT ObjectPtr<IPoolingLayer> SmartEngine::CreatePoolingLayer ( const PoolingLayerCInfo & cinfo )

Creates an instance of IPoolingLayer

◆ CreatePPOTrainer()

SMARTENGINE_EXPORT ObjectPtr<IPPOTrainer> SmartEngine::CreatePPOTrainer ( const PPOTrainerCInfo & cinfo )

Creates an instance of IPPOTrainer

◆ CreateStandardGraphModel()

SMARTENGINE_EXPORT ObjectPtr<IStandardGraphModel> SmartEngine::CreateStandardGraphModel ( const StandardGraphModelCInfo & cinfo )

Creates an instance of IStandardGraphModel.

◆ CreateStopGradientNode()

SMARTENGINE_EXPORT ObjectPtr<IGraphNode> SmartEngine::CreateStopGradientNode ( const StopGradientNodeCInfo & cinfo )

Stops the gradient from flowing backwards from this point. Only meaningful when training with gradient descent or reinforcement learning.

◆ CreateUnpoolingLayer()

SMARTENGINE_EXPORT ObjectPtr<IUnpoolingLayer> SmartEngine::CreateUnpoolingLayer ( const UnpoolingLayerCInfo & cinfo )

Creates an instance of IUnpoolingLayer

◆ CreateUserAgent()

SMARTENGINE_EXPORT ObjectPtr<IAgent> SmartEngine::CreateUserAgent ( const UserAgentCInfo & cinfo )

Creates an overridable instance of IAgent from a previously serialized agent.

◆ CreateUserGraphModel()

SMARTENGINE_EXPORT ObjectPtr<IGraphModel> SmartEngine::CreateUserGraphModel ( const UserGraphModelCInfo & cinfo )

Creates an overridable instance if IGraphModel

◆ DynamicCast() [1/2]

template<typename D >

ObjectPtr<D> SmartEngine::DynamicCast ( const IObject * p )

Safe cast an IObject to a pointer of a different type.

◆ DynamicCast() [2/2]

template<typename D >

ObjectPtr<D> SmartEngine::DynamicCast ( IObject * p )

Safe cast an IObject to a pointer of a different type.

◆ DynamicCastAttach()

template<typename D >

ObjectPtr<D> SmartEngine::DynamicCastAttach ( IObject * p )

Safe cast an IObject to a pointer of a different type. Does not call AddRef() on the interface.

◆ HashString()

SMARTENGINE_EXPORT HashKey SmartEngine::HashString ( const char * str )

◆ IsA()

template<typename D >

ObjectPtr<D> SmartEngine::IsA ( IObject * p )

◆ SetAssetStoreHandlers()

SMARTENGINE_EXPORT void SmartEngine::SetAssetStoreHandlers	(	AssetStoreHandler	loadHandler,
		AssetStoreHandler	saveHandler
	)

Used to set the handler for custom loading and saving of assets.

◆ SetLogHandler()

SMARTENGINE_EXPORT void SmartEngine::SetLogHandler ( LogHandler handler )

Used to set the handler for log output from the SmartEngine AI plugin

◆ SetSmartEngineLicense()

SMARTENGINE_EXPORT bool SmartEngine::SetSmartEngineLicense	(	const byte *	data,
		int64	count
	)

Loads a license from a license file blob. Call this before creating a context if you want to load a license from other than the default location.

Parameters

data

Variable Documentation

◆ cDefaultUpdateRate

const float SmartEngine::cDefaultUpdateRate = -2.0f

Use the model's default update rate. Not valid when registering a model. This is the only value that can be used to group controllers to the same network.

◆ cUpdateEveryFrameRate

const float SmartEngine::cUpdateEveryFrameRate = 0.0f

An update rate set to this value means we should update the controller with model output every frame.

◆ cUpdateManuallyRate

const float SmartEngine::cUpdateManuallyRate = -1.0f

An update rate set to this value means we shouldn't update the controller automatically, but rather only when UpdateControllerImmediately() is called.

Classes

Typedefs

Enumerations

Functions

Variables

Typedef Documentation

◆ Agent_EndEpisode_Ptr

◆ Agent_GiveReward_Ptr

◆ Agent_RecordAction_Ptr

◆ Agent_RecordObservation_Ptr

◆ Agent_StartNewExperience_Ptr

◆ AssetStoreHandler

◆ GraphModel_AddData_Ptr

◆ GraphModel_ClearData_Ptr

◆ GraphModel_Execute_Ptr

◆ GraphModel_RequestOutput_Ptr

◆ GraphModel_Reset_Ptr

◆ GraphModel_RetrieveOutput_Ptr

◆ HashKey

◆ MatrixIndex

◆ ObjectClassId

◆ ObjectId

Enumeration Type Documentation

◆ ActivationType

◆ AssetType

◆ ChoiceType

◆ Conv2DType

◆ DeviceType

◆ GeneSwapTarget

◆ GeneticAgentScoringMethod

◆ GradientDescentTrainingAlgorithm

◆ LayerType

◆ LogLevel

◆ LossTrainingMethod

◆ MutationTarget

◆ PoolType

◆ SerializationResult

Function Documentation

◆ CreateA2CTrainer()

◆ CreateActivationNode()

◆ CreateAgentDataStore()

◆ CreateBufferInput()

◆ CreateChoiceNode()

◆ CreateComponentInput()

◆ CreateConcatNode()

◆ CreateContext()

◆ CreateConv2DLayer()

◆ CreateCuriosityModule()

◆ CreateD4PGTrainer()

◆ CreateEmptyGraph()

◆ CreateExtractNode()

◆ CreateGeneticAgentTrainer()

◆ CreateGeneticTrainer()

◆ CreateGradientDescentTrainer()

◆ CreateGraphComponent()

◆ CreateGraphController()

◆ CreateGraphFromDefinition()

◆ CreateGraphFromResource()

◆ CreateGraphManager()

◆ CreateLoss()

◆ CreateMatrix()

◆ CreateMaximumNode()

◆ CreateMemoryBuffer()

◆ CreateMinimumNode()

◆ CreateMultiplexerNode()

◆ CreateNeuronLayer()

◆ CreateNormalNode()

◆ CreateParameter()

◆ CreatePoolingLayer()

◆ CreatePPOTrainer()

◆ CreateStandardGraphModel()

◆ CreateStopGradientNode()

◆ CreateUnpoolingLayer()

◆ CreateUserAgent()

◆ CreateUserGraphModel()

◆ DynamicCast() [1/2]

◆ DynamicCast() [2/2]

◆ DynamicCastAttach()

◆ HashString()

◆ IsA()