Struct MultiTaskNeuralNetwork 

pub struct MultiTaskNeuralNetwork<S = Untrained> { /* private fields */ }

Multi-Task Neural Network with Shared Representation Learning

This neural network implements multi-task learning where multiple related tasks share common representations in lower layers while having task-specific layers for final predictions. This approach allows for better generalization and improved performance when tasks are related.

Architecture

The network consists of:

• Shared layers: Learn common representations across all tasks
• Task-specific layers: Learn task-specific transformations
• Multiple outputs: One output per task

Examples

use sklears_multioutput::multitask::{MultiTaskNeuralNetwork, TaskBalancing};
use sklears_multioutput::activation::ActivationFunction;
use sklears_core::traits::{Predict, Fit};
// Use SciRS2-Core for arrays and random number generation (SciRS2 Policy)
use scirs2_core::ndarray::array;
use std::collections::HashMap;

let X = array![[1.0, 2.0], [2.0, 3.0], [3.0, 1.0], [4.0, 4.0]];
let mut tasks = HashMap::new();
tasks.insert("task1".to_string(), array![[0.5], [1.0], [1.5], [2.0]]); // Regression task
tasks.insert("task2".to_string(), array![[1.0], [0.0], [1.0], [0.0]]); // Classification task

let mt_net = MultiTaskNeuralNetwork::new()
    .shared_layers(vec![20, 10])
    .task_specific_layers(vec![5])
    .task_outputs(&[("task1", 1), ("task2", 1)])
    .shared_activation(ActivationFunction::ReLU)
    .learning_rate(0.01)
    .max_iter(1000)
    .task_weights(&[("task1", 1.0), ("task2", 0.8)])
    .random_state(Some(42));

Implementations

impl MultiTaskNeuralNetwork<Untrained>

pub fn new() -> Self

Create a new MultiTaskNeuralNetwork

pub fn shared_layers(self, sizes: Vec<usize>) -> Self

Set the sizes of shared representation layers

pub fn task_specific_layers(self, sizes: Vec<usize>) -> Self

Set the sizes of task-specific layers

pub fn task_outputs(self, tasks: &[(&str, usize)]) -> Self

Configure task outputs

pub fn task_loss_functions( self, loss_functions: &[(&str, LossFunction)], ) -> Self

Set loss functions for specific tasks

pub fn task_weights(self, weights: &[(&str, Float)]) -> Self

Set task weights for multi-task loss computation

pub fn shared_activation(self, activation: ActivationFunction) -> Self

Set activation function for shared layers

pub fn task_activation(self, activation: ActivationFunction) -> Self

Set activation function for task-specific layers

pub fn output_activations( self, activations: &[(&str, ActivationFunction)], ) -> Self

Set output activation functions for specific tasks

pub fn learning_rate(self, lr: Float) -> Self

Set learning rate

pub fn max_iter(self, max_iter: usize) -> Self

Set maximum number of iterations

pub fn tolerance(self, tolerance: Float) -> Self

Set convergence tolerance

pub fn random_state(self, seed: Option<u64>) -> Self

Set random state for reproducibility

pub fn alpha(self, alpha: Float) -> Self

Set L2 regularization strength

pub fn batch_size(self, batch_size: Option<usize>) -> Self

Set batch size for training

pub fn early_stopping(self, early_stopping: bool) -> Self

Enable/disable early stopping

pub fn validation_fraction(self, fraction: Float) -> Self

Set validation fraction for early stopping

pub fn task_balancing(self, strategy: TaskBalancing) -> Self

Set task balancing strategy

impl MultiTaskNeuralNetwork<MultiTaskNeuralNetworkTrained>

pub fn task_loss_curves(&self) -> &HashMap<String, Vec<Float>>

Get the loss curves for all tasks

pub fn combined_loss_curve(&self) -> &[Float]

Get the combined loss curve

pub fn n_iter(&self) -> usize

Get training iterations

pub fn task_outputs(&self) -> &HashMap<String, usize>

Get task configurations

Trait Implementations

impl<S: Clone> Clone for MultiTaskNeuralNetwork<S>

fn clone(&self) -> MultiTaskNeuralNetwork<S>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<S: Debug> Debug for MultiTaskNeuralNetwork<S>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for MultiTaskNeuralNetwork<Untrained>

fn default() -> Self

Returns the “default value” for a type.

impl Estimator for MultiTaskNeuralNetwork<Untrained>

type Config = ()

Configuration type for the estimator

type Error = SklearsError

Error type for the estimator

type Float = f64

The numeric type used by this estimator

fn config(&self) -> &Self::Config

Get estimator configuration

fn validate_config(&self) -> Result<(), SklearsError>

Validate estimator configuration with detailed error context

fn check_compatibility( &self, n_samples: usize, n_features: usize, ) -> Result<(), SklearsError>

Check if estimator is compatible with given data dimensions

fn metadata(&self) -> EstimatorMetadata

Get estimator metadata

impl Fit<ArrayBase<ViewRepr<&f64>, Dim<[usize; 2]>>, HashMap<String, ArrayBase<OwnedRepr<f64>, Dim<[usize; 2]>>>> for MultiTaskNeuralNetwork<Untrained>

type Fitted = MultiTaskNeuralNetwork<MultiTaskNeuralNetworkTrained>

The fitted model type

fn fit( self, x: &ArrayView2<'_, Float>, y: &HashMap<String, Array2<Float>>, ) -> SklResult<Self::Fitted>

Fit the model to the provided data with validation

fn fit_with_validation( self, x: &X, y: &Y, _x_val: Option<&X>, _y_val: Option<&Y>, ) -> Result<(Self::Fitted, FitMetrics), SklearsError>

where Self: Sized,

Fit with custom validation and early stopping

impl Predict<ArrayBase<ViewRepr<&f64>, Dim<[usize; 2]>>, HashMap<String, ArrayBase<OwnedRepr<f64>, Dim<[usize; 2]>>>> for MultiTaskNeuralNetwork<MultiTaskNeuralNetworkTrained>

fn predict( &self, X: &ArrayView2<'_, Float>, ) -> SklResult<HashMap<String, Array2<Float>>>

Make predictions on the provided data

fn predict_with_uncertainty( &self, x: &X, ) -> Result<(Output, UncertaintyMeasure), SklearsError>

Make predictions with confidence intervals