MultiTaskSplitter<T>
Models & Types · AiDotNet.Preprocessing.DataPreparation.Splitting.DomainSpecific
Multi-task splitter that ensures consistent splits across multiple related tasks.
Inheritance: DataSplitterBase<T> → MultiTaskSplitter<T>
For Beginners
In multi-task learning, we train a model on multiple related tasks simultaneously. It’s important that the same samples appear in train/test consistently across all tasks to fairly evaluate transfer learning benefits.
How It Works
Example: For a model predicting both “sentiment” and “topic” from text:
- Same documents should be in training for both tasks
- Same documents should be in testing for both tasks
When to Use:
- Multi-task neural networks
- Transfer learning experiments
- Multi-output regression/classification
Constructors
| Constructor | Summary |
|---|---|
MultiTaskSplitter(double testSize, int taskColumn, bool stratifyByTask, bool shuffle, int randomSeed) |
Creates a new multi-task splitter. |
Properties
| Property | Summary |
|---|---|
string Description { get; } |
Methods
| Method | Summary |
|---|---|
DataSplitResult<T> Split(Matrix<T> X, Vector<T> y) |