Table of Contents

Class CompressionOptimizerOptions

Namespace
AiDotNet.AutoML
Assembly
AiDotNet.dll

Configuration options for the compression optimizer.

public class CompressionOptimizerOptions
Inheritance
CompressionOptimizerOptions
Inherited Members

Properties

AccuracyWeight

Gets or sets the weight for accuracy in fitness calculation (default: 0.5).

public double AccuracyWeight { get; set; }

Property Value

double

CompressionWeight

Gets or sets the weight for compression ratio in fitness calculation (default: 0.3).

public double CompressionWeight { get; set; }

Property Value

double

IncludeEncoding

Gets or sets whether to include encoding techniques (default: true).

public bool IncludeEncoding { get; set; }

Property Value

bool

IncludeHybrid

Gets or sets whether to include hybrid techniques like Deep Compression (default: true).

public bool IncludeHybrid { get; set; }

Property Value

bool

IncludePruning

Gets or sets whether to include pruning techniques (default: true).

public bool IncludePruning { get; set; }

Property Value

bool

IncludeQuantization

Gets or sets whether to include quantization techniques (default: true).

public bool IncludeQuantization { get; set; }

Property Value

bool

MaxAccuracyLoss

Gets or sets the maximum acceptable accuracy loss as a fraction (default: 0.02 = 2%).

public double MaxAccuracyLoss { get; set; }

Property Value

double

MaxTrials

Gets or sets the maximum number of trials to run (default: 20).

public int MaxTrials { get; set; }

Property Value

int

MinCompressionRatio

Gets or sets the minimum acceptable compression ratio (default: 2.0).

public double MinCompressionRatio { get; set; }

Property Value

double

RandomSeed

Gets or sets the random seed for reproducibility (default: null for random).

public int? RandomSeed { get; set; }

Property Value

int?

SpeedWeight

Gets or sets the weight for inference speed in fitness calculation (default: 0.2).

public double SpeedWeight { get; set; }

Property Value

double