Class MANNOptions<T, TInput, TOutput>

Namespace

AiDotNet.MetaLearning.Options

Assembly

AiDotNet.dll

Configuration options for Memory-Augmented Neural Networks (MANN) algorithm.

public class MANNOptions<T, TInput, TOutput> : IMetaLearnerOptions<T>

Type Parameters

T

The numeric data type used for calculations (e.g., float, double).

TInput

The input data type (e.g., Matrix<T>, Tensor<T>).

TOutput

The output data type (e.g., Vector<T>, Tensor<T>).

Inheritance

object

MANNOptions<T, TInput, TOutput>

Implements

IMetaLearnerOptions<T>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Memory-Augmented Neural Networks combine a neural network controller with an external memory matrix. The network can read from and write to this memory, enabling rapid learning by storing new information directly in memory during adaptation.

For Beginners: MANN is like a neural network with a notebook:

1. The "controller" (neural network) processes inputs
2. The "memory" stores important information for later
3. Read heads retrieve relevant memories
4. Write heads store new information

This allows one-shot learning - see an example once, store it, use it later!

Constructors

MANNOptions(IFullModel<T, TInput, TOutput>)

Initializes a new instance of the MANNOptions class with the required meta-model.

public MANNOptions(IFullModel<T, TInput, TOutput> metaModel)

Parameters

metaModel IFullModel<T, TInput, TOutput>

The controller network to be trained (required).

Examples

// Create MANN with minimal configuration
var options = new MANNOptions<double, Tensor<double>, Tensor<double>>(myController);
var mann = new MANNAlgorithm<double, Tensor<double>, Tensor<double>>(options);

// Create MANN with custom memory configuration
var options = new MANNOptions<double, Tensor<double>, Tensor<double>>(myController)
{
    MemorySize = 256,
    MemoryKeySize = 128,
    NumReadHeads = 4
};

Exceptions

ArgumentNullException

Thrown when metaModel is null.

Properties

AdaptationSteps

Gets or sets the number of adaptation steps.

public int AdaptationSteps { get; set; }

Property Value

int

Default is 1 (MANN uses memory-based adaptation).

CheckpointFrequency

Gets or sets how often to save checkpoints.

public int CheckpointFrequency { get; set; }

Property Value

int

ClearMemoryBetweenTasks

Gets or sets whether to clear memory between tasks.

public bool ClearMemoryBetweenTasks { get; set; }

Property Value

bool

Default is false.

Remarks

For Beginners: If true, memory is partially cleared between tasks. If false, memory persists, enabling lifelong learning.

DataLoader

Gets or sets the episodic data loader for sampling tasks. Default: null (tasks must be provided manually to MetaTrain).

public IEpisodicDataLoader<T, TInput, TOutput>? DataLoader { get; set; }

Property Value

IEpisodicDataLoader<T, TInput, TOutput>

EnableCheckpointing

Gets or sets whether to save model checkpoints.

public bool EnableCheckpointing { get; set; }

Property Value

bool

EvaluationFrequency

Gets or sets how often to evaluate the meta-learner.

public int EvaluationFrequency { get; set; }

Property Value

int

EvaluationTasks

Gets or sets the number of tasks to use for evaluation.

public int EvaluationTasks { get; set; }

Property Value

int

GradientClipThreshold

Gets or sets the maximum gradient norm for gradient clipping.

public double? GradientClipThreshold { get; set; }

Property Value

double?

Default is 10.0.

InnerLearningRate

Gets or sets the learning rate for the inner loop (not used in MANN).

public double InnerLearningRate { get; set; }

Property Value

double

Default is 0.01.

Remarks

Note: MANN doesn't perform gradient-based inner loop adaptation. This value is kept for interface compatibility.

LossFunction

Gets or sets the loss function for training. Default: null (uses cross-entropy loss internally).

public ILossFunction<T>? LossFunction { get; set; }

Property Value

ILossFunction<T>

MemoryKeySize

Gets or sets the dimension of memory keys.

public int MemoryKeySize { get; set; }

Property Value

int

Default is 64.

Remarks

For Beginners: The size of the "index" used to find memories. Larger keys can encode more specific queries.

MemoryRegularization

Gets or sets the memory regularization strength.

public double MemoryRegularization { get; set; }

Property Value

double

Default is 0.0 (no regularization).

MemoryRetentionRatio

Gets or sets the ratio of memory to retain when clearing.

public double MemoryRetentionRatio { get; set; }

Property Value

double

Default is 0.5 (keep 50%).

MemorySize

Gets or sets the number of memory slots.

public int MemorySize { get; set; }

Property Value

int

Default is 128.

Remarks

For Beginners: How many "pages" in the memory notebook. More slots = more storage, but more computation.

MemoryUsageThreshold

Gets or sets the threshold for memory usage when consolidating.

public double MemoryUsageThreshold { get; set; }

Property Value

double

Default is 0.1.

MemoryValueSize

Gets or sets the dimension of memory values.

public int MemoryValueSize { get; set; }

Property Value

int

Default is 64.

Remarks

For Beginners: How much information each memory slot can hold.

MetaBatchSize

Gets or sets the number of tasks to sample per meta-training iteration.

public int MetaBatchSize { get; set; }

Property Value

int

Default is 4.

MetaModel

Gets or sets the meta-model (controller network) to be trained. This is the only required property.

public IFullModel<T, TInput, TOutput> MetaModel { get; set; }

Property Value

IFullModel<T, TInput, TOutput>

Remarks

The meta-model acts as the controller that processes inputs and generates keys for memory read/write operations.

For Beginners: This is the neural network that makes decisions about what to read from and write to memory.

MetaOptimizer

Gets or sets the optimizer for network updates. Default: null (uses built-in Adam optimizer with OuterLearningRate).

public IGradientBasedOptimizer<T, TInput, TOutput>? MetaOptimizer { get; set; }

Property Value

IGradientBasedOptimizer<T, TInput, TOutput>

NumClasses

Gets or sets the number of output classes.

public int NumClasses { get; set; }

Property Value

int

Default is 5.

NumMetaIterations

Gets or sets the total number of meta-training iterations.

public int NumMetaIterations { get; set; }

Property Value

int

Default is 1000.

NumReadHeads

Gets or sets the number of read heads.

public int NumReadHeads { get; set; }

Property Value

int

Default is 1.

Remarks

For Beginners: How many different ways to query memory simultaneously. Multiple heads can retrieve different types of information.

NumWriteHeads

Gets or sets the number of write heads.

public int NumWriteHeads { get; set; }

Property Value

int

Default is 1.

OuterLearningRate

Gets or sets the learning rate for the outer loop (controller training).

public double OuterLearningRate { get; set; }

Property Value

double

Default is 0.001.

RandomSeed

Gets or sets the random seed for reproducibility.

public int? RandomSeed { get; set; }

Property Value

int?

ReadHeadOptions

Gets or sets options for the read head configuration.

public object? ReadHeadOptions { get; set; }

Property Value

object

UseCommonPatternsInitialization

Gets or sets whether to initialize with common patterns.

public bool UseCommonPatternsInitialization { get; set; }

Property Value

bool

Default is false.

UseFirstOrder

Gets or sets whether to use first-order approximation.

public bool UseFirstOrder { get; set; }

Property Value

bool

Default is true since MANN doesn't use gradient-based inner loop.

UseHierarchicalMemory

Gets or sets whether to use hierarchical memory organization.

public bool UseHierarchicalMemory { get; set; }

Property Value

bool

Default is false.

UseMemoryConsolidation

Gets or sets whether to use memory consolidation.

public bool UseMemoryConsolidation { get; set; }

Property Value

bool

Default is false.

Remarks

Memory consolidation periodically removes rarely-used memories to make room for new information.

UseMemoryPreInitialization

Gets or sets whether to pre-initialize memory.

public bool UseMemoryPreInitialization { get; set; }

Property Value

bool

Default is false.

UseOutputSoftmax

Gets or sets whether to apply softmax to output.

public bool UseOutputSoftmax { get; set; }

Property Value

bool

Default is true.

UseValueProjection

Gets or sets whether to project values before storing.

public bool UseValueProjection { get; set; }

Property Value

bool

Default is false.

WriteHeadOptions

Gets or sets options for the write head configuration.

public object? WriteHeadOptions { get; set; }

Property Value

object

Methods

Clone()

Creates a deep copy of the MANN options.

public IMetaLearnerOptions<T> Clone()

Returns

IMetaLearnerOptions<T>

A new MANNOptions instance with the same configuration.

IsValid()

Validates that all MANN configuration options are properly set.

public bool IsValid()

Returns

bool

True if the configuration is valid; otherwise, false.