Class TabularAugmentationSettings
- Namespace
- AiDotNet.Augmentation
- Assembly
- AiDotNet.dll
Tabular-specific augmentation settings with industry-standard defaults.
public class TabularAugmentationSettings- Inheritance
-
TabularAugmentationSettings
- Inherited Members
Remarks
For Beginners: These settings control how tabular (spreadsheet-like) data is augmented. Useful for improving model generalization on structured data.
Properties
DropoutRate
Gets or sets the feature dropout rate.
public double DropoutRate { get; set; }Property Value
Remarks
Default: 0.1 (10% of features dropped)
EnableFeatureDropout
Gets or sets whether feature dropout is enabled.
public bool EnableFeatureDropout { get; set; }Property Value
Remarks
Default: false
Randomly zeros out features during training. Good regularizer.
EnableFeatureNoise
Gets or sets whether feature noise is enabled.
public bool EnableFeatureNoise { get; set; }Property Value
Remarks
Default: true
Adds small random noise to numerical features.
EnableMixUp
Gets or sets whether MixUp for tabular data is enabled.
public bool EnableMixUp { get; set; }Property Value
Remarks
Default: true
Blends feature vectors from different samples. Effective for tabular data.
EnableSmote
Gets or sets whether SMOTE is enabled.
public bool EnableSmote { get; set; }Property Value
Remarks
Default: false (only enable for imbalanced data)
Synthetic Minority Over-sampling Technique for class imbalance.
MixUpAlpha
Gets or sets the MixUp alpha parameter.
public double MixUpAlpha { get; set; }Property Value
Remarks
Default: 0.2
NoiseStdDev
Gets or sets the standard deviation of feature noise.
public double NoiseStdDev { get; set; }Property Value
Remarks
Default: 0.01
SmoteK
Gets or sets the number of nearest neighbors for SMOTE.
public int SmoteK { get; set; }Property Value
Remarks
Default: 5
Methods
GetConfiguration()
Gets the configuration as a dictionary.
public IDictionary<string, object> GetConfiguration()