Class AggregationStrategyBase<TModel, T>

Namespace

AiDotNet.FederatedLearning.Aggregators

Assembly

AiDotNet.dll

Base class for federated aggregation strategies.

public abstract class AggregationStrategyBase<TModel, T> : FederatedLearningComponentBase<T>, IAggregationStrategy<TModel>

Type Parameters

TModel

Model/update representation.

T

Numeric type.

Inheritance

object

FederatedLearningComponentBase<T>

AggregationStrategyBase<TModel, T>

Implements

IAggregationStrategy<TModel>

Derived

FedAvgFullModelAggregationStrategy<T, TInput, TOutput>

FedBNFullModelAggregationStrategy<T, TInput, TOutput>

FedProxFullModelAggregationStrategy<T, TInput, TOutput>

ParameterDictionaryAggregationStrategyBase<T>

RobustFullModelAggregationStrategyBase<T, TInput, TOutput>

Inherited Members

FederatedLearningComponentBase<T>.NumOps

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Methods

Aggregate(Dictionary<int, TModel>, Dictionary<int, double>)

Aggregates model updates from multiple clients into a single global model update.

public abstract TModel Aggregate(Dictionary<int, TModel> clientModels, Dictionary<int, double> clientWeights)

Parameters

clientModels Dictionary<int, TModel>

Dictionary mapping client IDs to their trained models.

clientWeights Dictionary<int, double>

Dictionary mapping client IDs to their aggregation weights (typically based on data size).

Returns

TModel

The aggregated global model.

Remarks

This method combines model updates from clients using the strategy's specific algorithm.

For Beginners: Aggregation is like combining multiple rough drafts of a document into one polished version that incorporates the best parts of each.

The aggregation process typically:

1. Takes model updates (weight changes) from each client
2. Considers the weight or importance of each client (based on data size, accuracy, etc.)
3. Combines these updates using the strategy's algorithm
4. Returns a single aggregated model that represents the collective improvement

For example with weighted averaging (FedAvg):

• Client 1 (1000 samples): model update A
• Client 2 (500 samples): model update B
• Client 3 (1500 samples): model update C
• Aggregated update = (1000A + 500B + 1500*C) / 3000

GetStrategyName()

Gets the name of the aggregation strategy.

public abstract string GetStrategyName()

Returns

string

A string describing the aggregation strategy (e.g., "FedAvg", "FedProx", "Krum").

Remarks

For Beginners: This helps identify which aggregation method is being used, useful for logging, debugging, and comparing different strategies.

GetTotalWeightOrThrow(Dictionary<int, double>, IEnumerable<int>, string)

protected static double GetTotalWeightOrThrow(Dictionary<int, double> clientWeights, IEnumerable<int> clientIds, string paramName)

Parameters

clientWeights Dictionary<int, double>

clientIds IEnumerable<int>

paramName string

Returns

double