LabelSpreading
API reference for LabelSpreading
SemiSupervised.LabelSpreading
interface LabelSpreadingOptions {
kernel?: 'rbf' | 'knn' | ((X: number[][], Y: number[][]) => number[][]);
gamma?: number;
nNeighbors?: number;
alpha?: number;
maxIter?: number;
tol?: number;
}
constructor(options: LabelSpreadingOptions = {})Label spreading assigns labels to unlabeled data using a normalized graph
and soft clamping controlled by alpha.
Methods
fit(trainX: number[][], trainY: number[]): voidpredict(testX: number[][]): number[]predictProba(testX: number[][]): number[][]
Example
const ls = new LabelSpreading();
ls.fit(trainX, trainY);
const preds = ls.predict(testX);