Class MCDropoutNeuralNetwork<T>
- Assembly
- AiDotNet.dll
Implements Monte Carlo Dropout for uncertainty estimation.
public class MCDropoutNeuralNetwork<T> : NeuralNetwork<T>, INeuralNetworkModel<T>, INeuralNetwork<T>, IFullModel<T, Tensor<T>, Tensor<T>>, IModel<Tensor<T>, Tensor<T>, ModelMetadata<T>>, IModelSerializer, ICheckpointableModel, IParameterizable<T, Tensor<T>, Tensor<T>>, IFeatureAware, IFeatureImportance<T>, ICloneable<IFullModel<T, Tensor<T>, Tensor<T>>>, IGradientComputable<T, Tensor<T>, Tensor<T>>, IJitCompilable<T>, IInterpretableModel<T>, IInputGradientComputable<T>, IDisposable, IUncertaintyEstimator<T>Type Parameters
TThe numeric type used for calculations (e.g., float, double).
- Inheritance
-
MCDropoutNeuralNetwork<T>
- Implements
- Inherited Members
- Extension Methods
Remarks
For Beginners: MC Dropout is the simplest way to add uncertainty estimation to existing neural networks.
The idea is straightforward:
- Add dropout layers to your network
- Keep dropout active even during prediction (normally it's turned off)
- Run multiple predictions with different random dropout patterns
- The variation in predictions tells you the uncertainty
This is much easier than full Bayesian neural networks but still provides useful uncertainty estimates. It's like getting a second (and third, and fourth...) opinion from slightly different versions of your model.
Constructors
MCDropoutNeuralNetwork(NeuralNetworkArchitecture<T>, int)
Initializes a new instance of the MCDropoutNeuralNetwork class.
public MCDropoutNeuralNetwork(NeuralNetworkArchitecture<T> architecture, int numSamples = 50)Parameters
architectureNeuralNetworkArchitecture<T>The network architecture (should include MC dropout layers).
numSamplesintNumber of forward passes for uncertainty estimation (default: 50).
Remarks
For Beginners: Make sure your architecture includes MCDropoutLayer instances. The more samples you use, the better the uncertainty estimate, but prediction becomes slower. 50 samples is a good default that balances accuracy and speed.
Methods
EstimateAleatoricUncertainty(Tensor<T>)
Estimates aleatoric uncertainty.
public Tensor<T> EstimateAleatoricUncertainty(Tensor<T> input)Parameters
inputTensor<T>The input tensor.
Returns
- Tensor<T>
The aleatoric uncertainty estimate.
Remarks
For Beginners: MC Dropout primarily captures epistemic uncertainty. Aleatoric uncertainty would require the network to explicitly model output variance, so this is a simplified estimate.
EstimateEpistemicUncertainty(Tensor<T>)
Estimates epistemic uncertainty.
public Tensor<T> EstimateEpistemicUncertainty(Tensor<T> input)Parameters
inputTensor<T>The input tensor.
Returns
- Tensor<T>
The epistemic uncertainty estimate.
Remarks
For Beginners: MC Dropout excels at capturing epistemic uncertainty, which represents the model's lack of knowledge about the correct prediction.
PredictWithUncertainty(Tensor<T>)
Predicts output with uncertainty estimates using MC dropout.
public UncertaintyPredictionResult<T, Tensor<T>> PredictWithUncertainty(Tensor<T> input)Parameters
inputTensor<T>The input tensor.
Returns
- UncertaintyPredictionResult<T, Tensor<T>>
A prediction result augmented with uncertainty information.