import { RegressorBase } from '../base';
import { Params } from '../base/estimator';
import { MLPProps } from './mlpBase';
export interface MLPRegressorProps extends MLPProps {
}
/**
 * Multi-layer perceptron regressor trained with minibatch backpropagation
 * (sklearn `MLPRegressor` parity for the 'adam' and 'sgd' solvers; the
 * lbfgs solver is not implemented).
 *
 * Identity output activation with squared loss (sklearn's
 * `squared_loss = mean((y - p)^2) / 2`); single target `y: number[]` only.
 * The L2 penalty `alpha` applies to weights only, never biases.
 *
 * Early stopping (`earlyStopping: true`) holds out the LAST
 * `validationFraction` of the rows as a plain deterministic split (sklearn
 * shuffles via train_test_split — documented deviation) and restores the
 * weights with the best validation R².
 */
export declare class MLPRegressor extends RegressorBase {
    private params;
    private fitted;
    private coefsState;
    private interceptsState;
    private lossCurveState;
    private nIterState;
    constructor(props?: MLPRegressorProps);
    getParams(): Params;
    /** Mean training loss recorded at the end of each epoch. */
    get lossCurve(): number[];
    /** Number of epochs actually run by the last fit. */
    get nIter(): number;
    fit(trainX: number[][], trainY: number[]): void;
    predict(testX: number[][]): number[];
}